Samples of alluvial gravel, cubitermes sp termite mound soil and clay soil were taken in the localities of Makoua and Ignié, in the Cuvette Centrale and Pool departments respectively, according to the geographical coordinates given in Table 1 .

In the following text, alluvial gravels will be designated by the two letters (Ag), clay soil by the letters (Cl) and cubitermes sp termite soil by the letters (Csp).

Figure 1 shows samples of alluvial gravel (Ag), whitish clay soil (Cl) and grey-black cubitermes sp termite soil (Csp).

Alluvial gravel (0/14) was taken from the Makoua river and transported to the civil engineering laboratory in Brazzaville. Before testing, the alluvial gravel was sampled to obtain a particle size of 0/14 mm. In this mechanical improvement, alluvial gravel is the main material, clay soil and cubitermes sp termite mound soil are two corrective materials to the alluvial gravel, used in the proportions (10%, 15%, 20%, 25%, 30%) and (15%, 20%, 25%, 30%, 35%), respectively.

In Figure 2 , the alluvial gravel (Ag)-clay soil (Cl) and alluvial gravel (Ag)-cubitermes sp termite mound soil (Csp) mixtures were obtained by mixing the materials in the proportions chosen for this purpose. Figure 3 shows the mechanical stabilization of mixtures, followed by laboratory tests.

In the laboratory, the analyses were carried out after passing the clay soil and the cubitermes sp termite mound soil through a 2 mm sieve. Pour la séparation des particules, deux types d’essais ont été réalisés: le tamisage pour les grains de taille $\varphi$ > 80 μm selon la norme NFP94-056 [14] et al. sédimentation pour les grains de diamètre $\varphi$ ≤ 80 μm selon la norme NF P94-057 [15] . The particle size fraction is deduced from the recommendations of the particle size nomograms, which consider clays as particles < 0.002 mm, silts 0.002 - 0.06 mm, sands 0.06 - 2 mm, gravels 2 - 20 mm and pebbles 20 - 200 mm. The particle sizes corresponding to D₁₀, D₃₀ and D₆₀ by sieving weight are deduced from the particle size curves. The uniformity coefficient C_u and the curvature coefficient C_c were used to characterize the granulometry of Ag (0/14)-Cl (0/1) and Ag (0/14)-Csp (0/2) mixtures defined in accordance with the formulae below:

$C_U=\frac{D_{60}}{D_{10}}$ (1)

$C_C=\frac{D_{30}}{D_{10}\ast D_{60}}$ (2)

D_x is the particle size corresponding to x % by weight of the sieve.

For each fine content, three distribution tests of the granulometric fractions are determined and the average of the three tests is taken and plotted on the graph.

The Atterberg limits are determined in accordance with standard NF P 94-051 [16] . The liquidity (LL) and plasticity (PL) limits are determined by the fraction of soil passing through a 0.40 mm sieve. The plasticity index (PI) is expressed by the following relationship:

$\text{PI}=\text{LL}-\text{PL}$ (3)

Measuring the methylene blue adsorption capacity of a soil involves measuring the quantity of methylene blue absorbed by the 0/5 mm fraction of the clay soil and the cubitermes sp termite mound soil. This test characterizes the clay content of a soil, a parameter directly linked to the specific surface area of the soil, which reflects the overall quantity and quality (activity) of the clay fraction. The methylene blue value of a soil (BVS) is determined in accordance with the standard NF P 94-068 [17] .

The maximum dry density and optimum moisture content were defined using the modified Proctor test, in accordance with standard NF P 94-093 [18] .

The sand equivalent assesses the proportion of fine elements contained in the soil, defined in accordance with standard NF P 18-598 [19] .

California Bearing Ratio (CBR) is a test used to determine the mechanical characteristics and compaction of materials in pavement layers. It measures the shear strength of the material and enables the bearing capacity of the material to be calculated, by estimating its resistance to punching. It is the essential parameter for geotechnical testing prior to construction and is defined in accordance with standard NF P 94-078 [20] . The static modulus of mixes has been defined by the relationship Est = 5CBR, provided that the CBR of all mixes is greater than 10 [21] .

The resistance of the aggregate to fragmentation under the action of traffic and of the aggregates in terms of hardness, resistance to abrasion and resistance to polishing are defined according to the Los Angeles test procedure based on standard NF P18-573 [22] .

The micro-Deval test determines the wear resistance of an aggregate sample. This wear resistance for certain rocks is not the same in dry conditions or in the presence of water. The test is defined according to the NF EN 1097-1 [23] . The flattening coefficient is one of the tests used to characterize the more or less massive shape of aggregates. Gravel grains that are closer to a sphere or cube are the best, and the test is defined according to the standard NF EN 933-3 [24] .

Origin Pro 2019b software was used to determine the coefficient of determination and the correlations between the intrinsic parameters of the mixtures. The mathematical model selected was the one with a coefficient of determination R² greater than 0.8. Minitab 17 software was used to determine the statistical properties of the granulometric fractions of the mixtures.

The particle size distribution of alluvial gravel (0/14) as the main material and silty clay (0/1) and clayey silt (0/2) as corrective materials extracted from Figure 4 are shown in Table 2 .

Cl-clay, Fsi-fine silt, Msi-medium silt, Csi-coarse silt, Fsa-fine sand, Msa-medium sand, Csa-coarse sand, Fg-fine gravel, Mg-medium gravel, Cg-coarse gravel, Csp-cubitermes sp, G-gravel.

According to Table 2 , alluvial gravel is a material with a good skeleton, lacking fine clay to ensure cohesion. Alluvial gravel is made up of smooth, whitish, rounded grains of size 0/14 [24] . The geotechnical properties of the alluvial gravel, the clay soil and the cubitermes sp termite mound soil are shown in Table 3 .

MDD (T/m³): maximum dry density, OMC (%): optimum moisture content, LA (%): Los Angeles, PI (%): plasticity index, LL (%): liquidity limit, PL (%): plasticity limit, SE (%): sand equivalent, MD: micro-Deval in the presence of water, UC: uniformity coefficient, CC: curvature coefficient.

According to the Unified Soil Classification System (USCS) [26] , clay soil (Cl) is classified as silty clay (Cl) and cubitermes sp termite mound soil (Csp) is classified as clayey silt (Csp). The uniformity and curvature coefficients of silty clay and clayey silt are not measurable [26] .

Figure 5 shows the particle size distribution of the Ag (0/14)-Cl (0/1) and Ag (0/14)-Csp (0/2) mixtures and the distribution of the grain size in the mixtures in accordance with standard NF EN 13,285 [25] and the technical document [21] .

In Figure 5 , the particle size curves between 0.08 and 2 mm remain spread out. In other words, the curves are oblique, meaning that the soils contain several classes of grain. These are good materials for building site tracks. The 30% silty clay (Cl) mixture does not integrate the grain distribution spindle defined by CEBTP 1984 [21] . Mixes with 15%, 20% and 25% Cl intermittently integrate the spindles. For the combination of alluvial gravels (Ag) and clayey silts (Csp), the 10% and 15% mixes incorporate both spindles for grain sizes below 6 mm. However, all the mixes incorporate the two spindles specified in European standard NF EN 13285 [25] for grain sizes below 7 mm. The geotechnical characteristics of the mixes are given in Table 4 and Table 5 .

MDD (T/m³): maximum dry density, OMC (%): optimum moisture content, CBR (%): California Bearing Ratio, SE (%): sand equivalent, Cu: coefficient of uniformity, Cc: coefficient of curvature.

The 10% mixture has a spread particle size distribution Cu (70.7) > 6, but poorly calibrated Cc (11.6) > 3 [26] . The uniformity and curvature coefficients of the 15%, 20%, 25% and 30% mixes cannot be determined because of the high proportion of fines (over 10%), making it impossible to measure the diameter corresponding to 10% of the passings in these mixes.

MDD (T/m³): maximum dry density, OMC (%): optimum moisture content, CBR (%): California Bearing Ratio, SE (%): sand equivalent, Cu: coefficient of uniformity, Cc: coefficient of curvature.

The 15% and 20% mixtures have spread out particle size distributions (Cu > 6) but are poorly calibrated (Cc > 3) [26] . The coefficients of uniformity and curvature of the curves for mixtures with 25%, 30% and 35% clayey silt cannot be determined because of the high proportion of fines (over 10%), making it impossible to measure the diameter corresponding to 10% of the passages in these mixtures. The fractions of fine sand, medium sand, coarse sand, fine gravel and medium gravel extracted from Figure 5 are shown in Figure 6 .

In Figure 6 , for the alluvial gravel (Ag)-silty clay (Cl) mixture, the relationship obtained between the distribution of particle fractions in the mixtures as a function of the addition of sandy clay are polynomial fits:

$Y_1=B+B_{1}X+B_{2}X$ (4)

From Table 6 , the coefficients of determination R² for the particle size fractions (medium sand and fine gravel) of 0.564 and 0.672 respectively, are less than 0.8. In Figure 6 , for the Alluvial Gravel (Ag)-Clayey silt (Csp) mixture, the relationship obtained between the distribution of particle fractions in the mixtures as a function of the addition of clay silt are polynomial fits:

$Y_2=B+B_{1}X+B_{2}X$ (5)

In Table 7 , the coefficients of determination R² for all the granulometric fractions are greater than 0.8.

According to Figure 7 , the 95% confidence intervals for the particle size fractions of the Ag-Cl mixtures, the confidence interval for the fine sand content has a mean of (19.336 - 26.696), a median of (17.335 - 29.143) and a standard deviation of (5.98 - 11.484). Medium sand has a mean (4.551 - 5.304), median (4.455 - 5.388) and standard deviation (0.611 - 1.174). Coarse sand has a mean (7.821 - 8.573), median (7.635 - 8.708) and standard deviation (0.611 - 1.174). Fine gravel has a mean (27,290 - 29,538), median (27,355 - 30,090) and standard deviation (1826 - 3507). Medium gravel has a mean (32.796 - 37.862), a median (30.090 - 37.325) and a standard deviation (4.116 - 7.904). The statistical properties of the particle size fractions extracted from Figure 7 are shown in Table 8 .

FSa: fine sand, MSa: medium sand, CSa: coarse sand, FG: fine gravel, MG: medium gravel.

According to Table 8 , the normality of MSa, CSa, FG and MG is less than 0.005, with variances of 0.6460, 0.6463, 5.767 and 29.286 respectively.

According to Figure 8 , the 95% confidence intervals for the particle size fractions of the Ag-Csp mixtures, the confidence interval for the fine sand content has a mean (15.146 - 19.734), a median of (15.850 - 21.82) and a standard deviation

(3.323 - 6.790). Medium sand has a mean (4.291 - 4.650), median (4.28 - 4.33) and standard deviation (0.260 - 0532). Coarse sand has a mean (10.002 - 12.313), median (8.496 - 12.47) and standard deviation (1.674 - 3.42). Fine gravel has a mean (33.107 - 33.408), median (33.252 - 33.45) and standard deviation (0.218 - 0.446). Medium gravel has a mean (32.401 - 34.784), median (31.852 - 33.880) and standard deviation (1.26 - 3.527). The statistical properties of the particle size fractions extracted from Figure 8 are shown in Table 9 .

FSa: fine sand, MSa: medium sand, CSa: coarse sand, FG: fine gravel, MG: medium gravel.

According to Table 9 , the normality of the mean sand of the clayey silt (MSa) and that of the fine gravels is less than 0.005, with respective variances of 0.122 and 0.086.

Figure 9 shows the evolution of the sand equivalent of the alluvial gravel as a function of the addition of silty clay fines (Cl) and clayey silt (Csp).

Figure 9 shows the decrease in the sand equivalent of the mixture as a function of the content of silty clay (Cl) and clayey silt (Csp). The sand equivalent of the mixture is not really proportional to the contents (Cl, Csp) of the mixture ( Figure 9 ). In fact, for silty clay (Cl), we have:

$SE=\left(89\pm 1.088\right)+\left(-0.52\pm 0.046\right){\alpha }_{Cl};R^2=0.970$ (6)

$\begin{array}{l}SE=\left(87.891\pm 0.676\right)+\left(-0.274\pm 0.079\right){\alpha }_{Cl}\\ \pm \left(-0.007\pm 0.002\right){\alpha }_{Cl}^2;R^2=0.993\end{array}$ (7)

For clayey silt (Csp)

$SE=\left(72.714\pm 10.369\right)+\left(-2.423\pm 0.535\right){\alpha }_{Csp};R^2=0.837$ (8)

$\begin{array}{l}SE=\left(85.690\pm 5.690\right)+\left(-5.537\pm 0.806\right){\alpha }_{Csp}\\ \pm \left(0.104\pm 0.026\right){\alpha }_{Csp}^2;R^2=0.975\end{array}$ (9)

where, SE: sand equivalent of the mixture, ${\alpha }_{Cl}$: silty clay content (Cl) and ${\alpha }_{Csp}$: clayey silt content (Csp) of the mixture.

The sand equivalent does not obey the law of mixtures unlike the sand content of the mixture. In other words, the sand equivalent is not proportional to the sand content of the mixture. In fact, the sand content of the mixture as a function of the silty clay content (Cl) and the clayey silt content (Csp) are given by the following relationships:

$P_S^m=P_S^{Ag}+\left(P_S^{Cl}-P_S^{Ag}\right){\alpha }_{Cl}$ (10)

$P_S^m=P_S^{Ag}+\left(P_S^{Csp}-P_S^{Ag}\right){\alpha }_{Csp}$ (11)

With: $P_S^m,P_S^{Ag},P_S^{Cl}$ et $P_S^{Csp}$ respectively the percentages of sand in the mixture, alluvial gravel (Ag), silty clay (Cl) and clayey silt (Csp); ${\alpha }_{Cl}$ et ${\alpha }_{Csp}$ respectively the silty clay content and the clayey silt content of the mixture.

If the sand equivalent of the mixture were proportional to the sand content of the mixture, it would also be proportional to the silty clay content (Cl) and the clayey silt content (Csp). Figure 10 shows the evolution of maximum dry density as a function of optimum moisture content.

According to Figure 10 , the maximum dry density decreases with increasing fine sand content and water content in the mixes ( Figure 6 ). The addition of silty clay and clayey silt increases the optimum moisture content of the mix.

Dry density is not a direct indication of a material’s mechanical strength. In a material that nevertheless has pores, the more interactions there are between the particles, the better the cohesion. The correlation between CBR index and maximum dry density as a function of silty clay (Cl) and clayey silt (Csp) is a polynomial fit:

$Y_3=B+B_{1}X+B_2{X}^2$ (12)

In Table 10 , the coefficients of determination R² for determining the correlation between the CBR and the maximum dry density of the blends are greater than 0.8.

According to Table 1 , the particle size distribution of alluvial gravel (0/14) is composed of sands (fine, medium, coarse) and gravels (fine, medium). Its particle size distribution is spread out and well calibrated, with uniformity coefficients Cu (9.2) > 4 and curvature coefficients CC (2.3) between 1 < Cc < 3 [26] . However, the uniformity and curvature coefficients for silty clay and clayey silt are not measurable. In other words, the particle size distributions of the two corrective materials are poorly calibrated. Its Los Angeles LA coefficient (41) is higher than the maximum of 35% required for the hardness of alluvial gravels to be used in the base layer of pavements [21] . Alluvial gravel integrates the two spindles in the range 2 - 8 mm and beyond that, the distribution of grains is outside the spindle.

According to the Unified Soil Classification System (USCS), the clayey soil and the soil of the cubitermes sp termite mounds are classified as highly plastic silty clay (Cl) and low plasticity clayey silt (Csp) respectively [26] . Mixe with Cl (10%) and Csp (15%) integrate the spindles and lie outside the two spindles for grain distribution greater than 8 mm ( Figure 5 ). Both mixes are suitable for road construction [21] . In fact, the dislocation of the grains obtained after compacting the material means that the grain distribution completely integrates the two spindles ( Figure 5 ). According to Figure 6 , the addition of silty clay (Cl) and clayey silt (Csp) to alluvial gravels increases the content of fine sand (FSa) in the mixture. The distribution of Cl (coarse sand, medium gravel) and Csp (fine gravel) grains varies little. The distribution of Cl and Csp grains (coarse sand, medium gravel) and Csp (medium sand) decreases ( Figure 6 ).

In Table 6 , Table 7 and Table 10 , the equations used are those with coefficients of determination R² ≥ 0.8.

The maximum CBR values (96%, 84%) Figure 11 were obtained with Cl (10%) and Csp (15%) mixes, with sand equivalents of 35% ( Table 4 ) and 82% ( Table 5 ) respectively. All these values are within the authorised limits for road construction. Despite a CBR > 80% (CEBTP 1984) [21] , mixtures with Cl (10%) and Csp (15%) cannot be used as base layer materials for pavements. In fact, alluvial gravel with a grain size (0/14) and LA value (41%) > 35% (CEBTP 1984) [21] is not suitable as a support for hammering tyres in a pavement base layer.

The addition of silty clay (Cl) and clayey silt (Csp) increases the fine sand content of the mix, which reduces the maximum dry density ( Figure 10 ) and the CBR index ( Table 4 and Table 5 ). The increase in optimum moisture content is a function of the increase in Cl and Csp fines in the mix ( Figure 10 ). The coefficients of determination as a function of grain distribution in the Ag-Csp mixtures (0.839 - 0.996) are greater than 0.8 ( Figure 12 ).

In Figure 12 , the coefficients of determination as a function of grain distribution in Ag-Cl mixtures for MSa and FG particles are less than 0.8 and those for FSa, CSa and MG particles are greater than 0.8.

According to Figure 7 and Table 5 , the Anderson-Darling normality test with the P value of the distribution of grains in the granulometric fractions (Msa, Csa, FG, MG), are less than 0.005, with A2 (1.15 - 1.69) and that of Fsa (0.042) with A2 (0.75) and the variance (0.646 - 61.823) by the addition of silty clay fines. According to Figure 8 and Table 6 , the addition of clayey silt (Csp), the Msa and FG particle size fraction has a P value of less than 0.005, with A2 (2.91 - 2.29) respectively, and that of FSa (0.095), CSa (0.025), MG (0.021) has A2 (0.61 - 0.86) with a variance (5.050 - 19.902). The results obtained for each type of mixture (Ag-Cl and Ag-Cs) are disparate, which may be linked to the law of mixtures. Indeed, according to Figure 9 , the sand equivalent does not seem to obey the law of mixtures, unlike the sand content of the mixture. In other words, the sand equivalent is not proportional to the sand content of the mixture. If the sand equivalent of the mixture were proportional to the sand content of the mixture, it would also be proportional to the silty clay content (Cl) and the silty clay content (Csp). All these changes can be explained by the fact that the distribution of grains in the mixtures may not obey the law of mixtures [27] [28] .