The rate of deterioration of gravel base course materials may influence the durability of flexible pavements. In particular, selected gradation curves usually satisfy filter criteria thus avoiding migration of fines from one layer (subgrade, base, and subbase) into adjacent one. This paper shows that even if filter criteria are met between successive layers, some 0–20 mm materals do not satisfy the fines mobility criteria that presently are not considered in gradation criteria. This means that fines (<80 μm) can move within the pores of the layer. Three gradations of two 0–20 mm crushed stones were tested in the laboratory. Experiemrnts were designed to avoid degradation (breakage of particles), in order to study segregation only. One of the tested materials (No. 2) satisfies the gradation criteria but does not satisfy the fines mobility criteria. The experimental results confirmed that material No. 2 is subject to marked segregation during dry placement and compaction. When palced moist, it has a negligible initial segregation, but cyclic loading produces segregation. This internal instability was predictable from fines mobility (suffossion) criteria. It produces layers with different fines contents, which would reduce vertical drainage, increase pore pressures produced by traffic loading, and increase both capillary retention and frost damage. Key words: road, gravel base course, segregation, suffossion, seepage, gradation.

New laboratory results are presented on how densification influences the hydraulic anisotropy of a sand. Results are given for two compaction modes. Tests were performed at different densities in order to derive the curves kv(e) and kh(e). The samples were statically compacted in the first series and dynamically compacted in the second series. The ratio rk = kh/kv is different for the two series. It is higher than 1 for static compaction and lower than 1 for dynamic compaction. Thus, the anisotropy ratio of granular soils can take several values for the same void ratio, depending on the preparation mode. Even if the anisotropy ratios differ, the first invariant of the hydraulic conductivity tensor, 11k, is found to be a function of e only for this sand, regardless of the densification mode. The evolution of an adimensional anisotropy index, I1k/kv, versus a density index, Ie, is compared for the sand and a dispersed clay. The evolutions are nearly identical, which means that the hydraulic anisotropy of natural homogeneous soils may be similar for sediments having settled in still water and influenced subsequently by gravity only. From an engineering viewpoint, a reasonably good fit is obtained between experimental results and predictive charts. The results also demonstrate that the high apparent anisotropy often found either in natural or in compacted soils is not induced mainly by deposition or compaction but rather is due to nonuniformity or segregation in the soil mass. Key words: soil, permeability, anisotropy, laboratory, compaction.

The standard test method used to determine the hydraulic conductivity of a clean granular soil (rigid-wall permeameterconstant head difference) requires that the specimen be saturated with an air-vacuum pump. However, no method is provided to verify whether the sample is fully (100%) saturated. This paper proposes such a method and establishes its accuracy. The method allows for quality control of the rigid-wall permeameter itself: it can detect if it is watertight but not airtight, according to the achieved degree of saturation. A detailed example is given on a proposed data sheet. The reasons for partial saturation and the problems related to air bubbles are examined. The result of this examination is a simplified test to check a permeameter, based on the fact that for unsaturated conditions the hydraulic conductivity depends on which of two ways the water seeps in a given direction. A few modifications in the preparation of the specimen and the test procedure are suggested so as to improve the quality of test results.Selon la norme relative a la mesure de la conductivitC hydraulique des sols granulaires propres (permkamhtre a paroi rigide -essai a difference de charge constante), 1'Cchantillon doit &tre saturC a l'aide d'une pompe a vide. La norme ne propose toutefois pas de mCthode pour vCrifier si 1'Cchantillon est saturC a 100% ou non. Le prCsent article propose une mCthode dont il Ctablit la pricision. Cette mCthode permet de contr6ler la qualitC du permiamhire paroi rigide : on pourra savoir s'il est Ctanche a l'eau mais pas a l'air, selon le degrC de saturation atteint. Un exemple dCtaillC est trait6 sur une feuille d'essai proposCe. Les raisons d'une saturation partielle et les problhmes reliCs aux bulles d'air sont discutb. 11 en rCsulte un test simplifik pour vCrifier le permkamhtre, bask sur le fait que la conductivitC hydraulique, en rtgime non saturi, dCpend du sens de 1'Ccoulement dans une direction donnte. Quelques modifications dans la preparation de 1'Cchantillon et la procedure d'essai sont suggCrCes afin d'amCliorer la qualit6 des rCsultats.