Representative culms of Bambusa vulgaris were obtained from a four year old bamboo stand at Nteje (6˚16"N latitude and 6˚40"E longitude) in Oyi Local Government Area, Anambra State, southeast, Nigeria. The culms were collected at 10, 20, 50, 70 and 90 percent of the full length for each stick of bamboo. Identification was done using floral parts after the method reported by some authors [22] [23] . The collected culms were allowed to dry in the open laboratory for 30 days at the end of which each culm was manually broken into 100 mm by 150 mm by wall thickness. The resultant discs were further reduced to flakes of average 50 mm by 16 mm by 1.9 mm in a pallmann knife ring flaker. The flakes were further allowed to dry in the open laboratory for two weeks at laboratory conditions. At the end of this period, the flakes were collected and stored in vapour impermeable bags and kept for use. Cement used was ordinary Portland cement sold in 50 kg bags and commonly available in Nigerian market.

Representative samples of the flakes were either marked and used untreated, to be used after soaking in distilled water for 24 hrs at room conditions; soaked in distilled water at a temperature of 80˚C or soaked in one percent Sodium Hydroxide solution (NaOH) for 24 hrs at room conditions. At the end of each soak treatment, the leachants were allowed to drain off and the flakes washed immediately with copious supply of distilled water. The soaked and unsoaked particles were conditioned in a conditioning chamber at 20˚C ± 2˚C and 65% ± 1% Relative humidity (R. H.) and stored in vapour impermeable bags ready for use.

Proximate chemical analysis was conducted in compliance with the provisions of the standard Test Methods of the Technical Association of the Pulp and Paper Industry [24] . Representative flakes randomly selected were Wiley milled to powder, screened through a 60 BSS mesh and powder that passed through 60 BSS mesh but retained on 80 were used for the analysis.

Representative flakes were selected and reduced to splinters. The splinters were macerated/pulped in 1% NaOH conc. at 80˚C for 10 hrs. At the end of the above period, the resultant fiber bundles were washed in distilled water and the fiber suspension subsequently subjected to further defiberization in the presence of glass beads with the aid of vigorous agitation (Onuorah, 2005). One hundred and thirty (130) representative fiber samples were selected, stained with safarin-C and mounted on slides and viewed at 100× magnifications using a microscope. The fiber dimensions (length, diameter, lumen width and cell wall thickness) were measured wet with the aid of stage and occular micrometer scales [15] [25] . The averages and standard deviations were determined.

Thirty representative culms were taken and the culm lengths, culm diameter and culm wall thickness measured using linear tapes and micro gauge. The specific gravity was determined in accordance with the provisions of ASTM D2395-89 [26] . The means and standard deviations were determined and recorded.

Appropriate quantity (oven dry weight basis) of the B. vulgaris flakes for each soak treatment type was measured out based on cement-bamboo flakes mix ratio. The exact quantity to be used for each board type was determined in preliminary studies which took into cognizance the soak treatment type/mix ratio/composite density combinations. The required quantity of water in litres (Q) needed for the production of each board was determined using Equation (1)

where C = cement wt in kg; M = fiber/flake moisture content and W = oven dry wt of fiber/flakes as reported by Erakhrumen [7] . Production variables used were as in Table 1.

The production constants were as follows: (all pressing was done cold i.e. absence of external heat)

1) Prepressing pressure of 0.5 N/mm² before final consolidation.

2) Final press pressure of 1.4 N/mm² for 24 hrs.

3) Curing of the pressed composite took place inside a vapour impermeable bag over a 28-day period.

4) Use of 6 mm thick metal stops placed on opposite sides of the metal caul/formed furnish.

5) Addition of 3% of calcium chloride (wt/wt cement basis) to weighed flakes before addition of cement.

6) Use of 19 mm plywood covered with thin polythene as the top platen.

Mixed furnish for each composite panel was formed on top of a metal cual whose surfaces had been previously covered with a thin polythene sheet while the top of the formed furnish was covered with 19 mm plywood whose surface had been covered with same polythene. Opposite each side of the length of the caul carry-

^kTwo composite panels were made per production mix from which two test specimens per panel were sampled for testing. Control = unsoaked flakes; cold water = unheated distilled water used for soaking the flakes for 24 hrs at room conditions; hot water = flakes soaked in hot water at 80˚C for one hour; 1% NaOH = flakes soaked in 1% concentration of NaOH for 24 hrs at room conditions; ^aExcept for the control all soak treatments were immediately followed by rapid but instantaneous washing in water in order to remove leachants and/or unspent chemicals.

ing the furnish but besides the furnish was placed a 6 mm metal stop. At the end of the curing period the panels were trimmed 0.5 mm from the edges in order to remove areas that were not properly consolidated during pressing.

The composite panels were sampled and tested in accordance with the provisions of ASTM D1037-96 [26] . The location of each test specimen in each composite panel was determined with the aid of a table of random numbers. All test specimens were conditioned in a controlled chamber at 20˚C ± 2˚C and 65% ± 1%. Relative humidity (R.H.) was for two weeks before testing.

Properties tested were the Modulus of Rupture (MOR) and Modulus of Elasticity (MOE) in Bending; the water absorption (WA) and thickness swelling (TS) due to a 24 hours horizontal water soak. There were a total of 4Nos tests for each production mix thus making a total of 160 Nos tests for each property. The reported properly value represents weighted average of four test samples at each production mix (mix ratio/soak/density interaction).

A complete randomized experiment was designed at which the effects of the variable production mix (5 Nos mix-ratios, 4 Nos soak types and 2 Nos composite densities) and all possible interactions were determined. The levels of significant effects of the main variables, their first order interaction and second order interaction were determined using factorial analysis [27] . The mean property values at appropriate production mix were also determined.

Table 2 presents the fiber morphological characteristics, the culm features and physical characteristics of the mature B. vulgaris grown in Nteje, lower Anambra river basin, southeast, Nigeria along side of similar features for bambooes globally, sugarcane bagasse, decidous hardwoods and coconut coir fiber globally as reported in the literature [28] [29] . These are the competing woody and non-woody fiber sources for composites. The fiber

(1) Fiber Dimensions are the weighted average of 130 fiber measurements. (2) Information on deciduous words and other non wood species used in this table are from Hunter (2001) and Asasutjarit (2005).

morphological properties and the density of the bamboo compare favourably with those of other woody and non-wood species used in the production of wood-cement composites in other countries. Of particular interest is the fiber slenderness ratio of 160.95:1 which is within the range of 135 to 175:1 reported for all species of bamboo [29] and found to be greater than 85:1 for sugar-cane bagasse; 50:1 for deciduous Hardwood species and 58.8:1 for matured coconut coir fiber [28] [29] . High slenderness ratio of fibers/particles has been accepted to enhance the binder utilization efficiency of wood-based composites with attendant property improvements [30] .

The proximate chemical analysis results are also in Table 2 and are also presented along side those for bambooes, deciduous hardwood species, sugar cane bagasse and coconut coir, all globally. The chemical constituents as well as leachable components compare favourable with those of bambooes globally as well as other com- peting woody and non-woody fiber sources used for composite manufacture globally [6] [15] [28] [31] . Chew and his co-workers reported that the effects of soluble wood constituents on cement hydration and consequently on wood-cement composite panel properties can effectively be reduced by reducing the soluble sugars to as low as 0.28% [12] . One other effective means of reducing the effects of soluble constituents on hydration is the use of additives. Available records indicate that incorporating CaCl₂ above 3% (cement wt basis) in furnish mixtures above 2:1 cement wood ratio (wt/wt—on oven dry wt wood basis) maybe adequate in creating optimal hydration conditions for effective bonding and optimal composite panel properties [5] [20] [32] .

Table 3 presents the results of Modulus of Rupture (MOR) and Modulus of Elasticity (MOE) in bending as well as the water absorption (WA) and thickness swelling (TS) due to a 24 hrs horizontal water soak. Of the eighty (80 Nos) composite panels produced in this study representing all possible combination of production mix, the average property ranges from 25.0 to 73.45 N/mm² for MOR; 4128 to 15,065 N/mm² for MOE; 15.01% to 36.11% for WA; and 3.04% to 12.72% for TS due to a 24-hr horizontal water soak. All properties met the specifications of American National Standard Institute—ANSI A208-1 and ANSI A 208-2; British standard specifications (BSS) [33] -[35] for wood chipboard and methods of test for particleboard BS5669 [35] and Malaysian Standard Specifications for wood cement Board MS 934 [36] . Table 4 presents the summary of factorial analysis on the effects of each manufacturing variable and all possible interactions of the variables on composite panel properties.

At any given cement—B. vulgaris mix ratio (1:1, 1.5:1, 2:1, 2.5:1 and 3:1 wt/wt) and soak treatment (control, cold water, hot water and 1% concentration of NaOH) the MOR increased as the ratio of the cement in the composite was increased (Table 3). This is in accord with the findings of other workers [8] [12] -[14] [32] . The effect of cement—B. vulgaris mix ratio on MOR was found significant at 5% level of probability in a factorial analysis while the effects of mix ratio at all possible levels of interactions of production mix used in the study were significant at 0.01% level. A possible explanation is that increase in composite binder content ensured more effective coverage of the surface area of the flakes and thus enhanced a strong inter fiber/flake bond until an optimum binder content will be reached beyond which increase in binder content will no longer bring additional enhancement in properties [8] [30] . Lingfei found optimal cement wood mix ratio to be 2.65:1 [20] . At any given cement—wood mix ratio and density, MOR was relatively enhanced as the soak treatment was changed from untreated/control to cold water, to hot water and to 1% concentration of NaOH treatments in that order (see Table 3). This is in agreement with the findings of other workers [8] [37] . The differences due to the effects of soak treatments were significant at 0.05% level but the effects of soak treatment interactions at all possible levels with other production mix were significant at 0.01% level (Table 4). The result is in conformity with known behaviour of wood-cement composites. It is generally understood that soluble sugars and extraneous materials interfer with cement hydration and consequently influence properties of composites. One accepted method of reducing the effects of these inhibitory constituents on cement hydration is the various soak treatment methods [10] [12] [20] [37] .

The effects of Board density on the MOR of the composite panel was not found significant at 0.05% level of probability in a factorial analysis. This was not expected as high density implies higher compaction ratios for each given production mix (mix ratio and soak type). However, it is widely acknowledged that there is optimal compaction ratio above which further improvements in wood based composite properties as a result of compaction ratio no longer results in significant properties improvements [30] . The exact optimum for wood-cement boards found in the literature varied with the choice of lignocellulosic furnish, mix ratio, soak treatment, cement/water mix ratio, additive type and quantity, etc. But there appears to be a synergy between panel density and all other production mix at all possible levels of interactions (density/soak treatment, density/cement: wood mix ratio; density/soak type/cement: wood mix ratio) as their effects were all found significant at 0.01% level. The explanation may be that above 800 kg/m³ effect due to soak type and/or cement: B. vulgaris mix ratio far out-weighs the effects due to density alone. It may also mean that differences in relative volume of the B. vulgaris flakes used in making the boards at all cement: wood ratios was insignificant taking into cognizance the weight of cement and average mean density of the bamboo relative to panel density.

At any given composite panel density and cement: B. vulgaris flakes mix ratio, the composite MOE in bending increased as the soak treatment was changed from unsoaked flakes/control to cold water soak to Hot water soak to 1% concentration of NaOH soak in that order (Table 3). At same level of soak treatment, the composite MOE increased as mix ratio was increased from 1:1 to 3:1 (Table 3). These are in agreement with the findings of other workers who used different lignocellulosic furnish [7] [8] [13] [14] . The effects of each production variable (density/soak treatment/cement: wood mix ratio) as well as all their possible interactions on MOE were all found

Reported values are the weighted average of four (4) tests per production mix. ( ) = Nos in parenthesis are the co-efficient of variations in %. All test specimens were conditioned at 20˚C ± 2˚C and 65% ± 1% relative humidity for 2 weeks before test.

^NSNot significant at 0.05 level (F_Tab = 3.91). ^*Significant at 0.05 level. ^**Significant at 0.01 level. ^*kF_Tab = 2.43 at 5% level of significance. ^*F_Tab = 2.67 at 5% level of significance.

significant at 0.01% level of probability. This tends to suggest that MOE is a more sensitive measure of the importance of manufacturing variables on properties of wood/cement composites than MOR. It also goes to suggest that structural integrity as could be adjudged from compaction ratio and/or relative void volume play more significant roles in determining MOE of wood-cement composite panels. This later argument is in accord with the findings of Lingfei who corroborated properties with internal structures of bamboo/cement composite panels using X-ray diffraction (XRD), Thermo-gravimetric differential analysis (TG-DTA) and Scanning Election Microscopy (SEM) [20] .