ROWE, P. W. (1969). Gdolechnique 19, No. 1, 75-86. Relation between the peak Coulomb 4 values for saturated drained sands measured in the direct shear test and the plane strain compression test. Present available experimental data indicate quite close agreement over the range +, = 17"-39" for cohesionless soils.
ROWE, P. W. (1969). Gdolechnique 19, No. 1, 75-86. Relation between the peak Coulomb 4 values for saturated drained sands measured in the direct shear test and the plane strain compression test. Present available experimental data indicate quite close agreement over the range +, = 17"-39" for cohesionless soils.
ROWE, P. W. (1969). Gdolechnique 19, No. 1, 75-86. Relation between the peak Coulomb 4 values for saturated drained sands measured in the direct shear test and the plane strain compression test. Present available experimental data indicate quite close agreement over the range +, = 17"-39" for cohesionless soils.
ROWE, P. W. (1969). Gdolechnique 19, No. 1, 75-86.
THE RELATION BETWEEN THE SHEAR STRENGTH OF SANDS
IN TRIAXIAL COMPRESSION, PLANE STRAIN AND DIRECT SHEAR P. IV. ROWE* SYNOPSIS A theoretical relation is derived between the peak Coulomb 4 values for saturated drained sands measured in the direct shear test and the plane strain compression test using the stress-dilatancy equation and the assumption that the directions of principal strain increment and principal stress coincide. Limiting dilatancy rates in triaxial compression allow an overall comparison of the Q values to be expected in these three types of test. Present available experimental data indicate quite close agreement over the range +, = 17-39 for cohesion- less soils. On derive une relation theorique entre les valeurs 4 de pointe de Coulomb pour des sables drain& sa- tures mesurCes dans lessai de cisaillement direct et lessai de compression d deformation en plan en util- isant 16quation de contrainte-dilatabilitk et la sup- position que les directions de laccroissement de dhformation principale et de la contrainte principale coincident. Des taux limites de dilatabilitb en com- pression tr axiale permettent de faire une comparai- son densemble des valeurs 4 auxquelles on peut satten.lre pour ces trois types dessais. Les don&es exp6rimentales disponibles actuelles indiquent une concordance t&s proche dans le cas oh la gamme des valeurs de 4, = 17-39 pour les sols sans coh&ion. INTRODUCTION The number of variables governing the shear strength of sands is so great that any one report must necessarily be confined to a limited aspect of the subject. In order that the many contributions from workers in different countries using different sands, apparatus, and technique may be related it is necessary to separate the strength component of particle struc- ture from that of inter-particle friction and to relate the strengths derived by means of a variety of stress systems. The present contribution derives the relation between the strength of a given sand in the plane strain and direct shear tests, expressed in terms of the peak effective stress ratio, and compares theoretical and experimental results with strength limits previously derived for triaxial compression. It is convenient first to review briefly the essential general findings which comprise the stress-dilatancy treatment of cohesionless soils subject to effective stress, when applied to the special boundary conditions of the triaxial test where either (TV >(TV =03 or q1 =(T~>(TV and to plane strain compression where o1 >u2 >o3 and e2 =O. The stress-dilatancy equation is the name given to one particular equation in this treat- ment, namely in compression R=DK . . . . . . . . (la) in extension R =K/D . . . . . . . . . (lb) where R=o;/oi, D=(l-dvs/dca,), vs is the volume decrease per unit volume, Ebb is the major principal compressive strain <I in a compression test and the minor principal compressive strain c3 in the extension test, due to interparticle slips, and K=tan2(35+$,/ 2). The value of & depends on the relative density, pressure range and the stress path and varies between a lower value &, being the angle of friction between the mineral particles, and an upper value d,, at the critical state. The conditions associated with the limit values of & are discussed later. * University of Manchester. 75 76 P. W. ROWE These relations which apply to the components of strain increments associated with slip movements, at all stages of deformation to failure, have been derived in three separate ways by Rowe (1962), Rowe, Barden and Lee (1964) and Horne (1965). A fuller treatment of the original derivation was reported by Barden and Khayatt (1966). The measured ratio &/de, of total volumetric increment dv and major principal strain increment de, includes elastic type strain increments prior to failure (Rowe, Barden and Lee, 1964) yet satisfies equation (1) closely over the major part of a stress path to failure in tests with increasing R at constant U& as reported by Rowe (1962), Lee (1966), Barden and Khayatt (1966) and Parikh (1967). Where D=l prior to peak, the experimental observations by Kirkpatrick (1961) also give +r values in agreement with the range described below. In the case of stress paths at constant R, separation of the elastic component from the slip component has led to equation (1) for the slip components (El-Sohby, 1964). Constant volume tests which exhibit constant R over a major portion of the stress path include marked pre-peak elastic components. It may be noted that although the term elastic refers to strain components associated with stored energy in grain compression, it may not necessarily be recoverable as with a true elastic component because some of the stored energy is absorbed in slips during unloading. This is particularly marked during unloading in varying R tests where repeated loading and unloading result in a continual absorption of energy in friction at interparticle slips. When considering soils at the peak, denoted by the maximum principal effective stress ratio, or at the critical state when after large strains no further change in stress ratio occurs with strain, small strain increments take place at constant effective stress and the elastic incremental strains are zero. In the case of plane strain compression, writing de, =de 2, =0, equations (1) apply and are identical, although the boundary strain conditions influence the value of 4r. Experimental verification of equation (1) has been reported by Rowe (1964a) and Wightman (1967). Procter (1967) found similar behaviour in triaxial compression, extension and plane strain conditions using the hollow cylinder test. The data have been based on measurements of the overall change in dimensions of a sand element rather than on internal strains but all these experiments have been conducted since 1953 using lubricated ends (Rowe, 1962). Kirkpatrick and Belshaw (1968) have reported evi- dence that the internal strains are sensibly uniform throughout the right cylinder, maintained during triaxial compression with lubricated ends. Cole (1967) and Roscoe (1967) report good agreement with the stress-dilatancy equation on the basis of more than 170 tests with Leighton Buzzard sand in the Cambridge Mk. 6 S.S.A. following important developments in technique which allowed the determination of principal stress and strain-rate ratios within a uniform central element. The present data confirm the following summary of statements concerning the approxi- mate value of & in equation (1) in compression (Rowe, 1962, 1963, 1964a, 1964b) and extension (Barden and Khayatt, 1966). In triaxial strain, compression or extension (b, I +r 5 do7 . f . . . . . . (2) In the densest state up to peak stress ratio #r=r$rr . . . . . . . . . . (3) In the loosest state at the peak stress ratio, which is the critical state r#f = +,, . . . . . . . . . . (4) In all cases where &<& at peak, namely in tests on sand other than in the loosest state, & increases to & from peak to the critical state. SHEARSTRENGTHOFSANDS 77 In $lane strain, for any packing up to the peak stress ratio, in compression or extension #I = +,, . . . . . . . . . . (5) Looking at the range of $t values (equation (2)) it is noted that the insertion of &=&., leads to a maximum value of K (equation (I)), namely K,,, in contrast to the insertion of $f=+LI giving a minimum value of K, namely K,. This maximum-minimum range of K values is, however, not to be confused with the minimum energy principle underlying the derivation of equation (l), nor need there be any confusion between the minimum energy principle and a minimum energy ratio. Equation (1) states the minimum absolute energy absorbed in interparticle friction at a given applied stress level and energy input and with respect to all possible instantaneous mean particle slip directions and not with respect to all possible $r values. In the latter connexion it may be noted that for the particular case of dv, =0 at the critical state, D =I and equation (1) is identical to the Rankine equation with +=& =&,. In the derivation of the Rankine equation the minimum energy principle is seen even more clearly if the Rankine equation is derived using Coulombs method of determining the mini- mum boundary force for all possible straight slip plane directions as follows. From Fig. 1 whence alb o;b cot /3 = tan ($+B) The minimum value of U; for a given uj is given when ,6=45-$12, and the insertion of this value of p into equation (6) gives the Rankine equation u;/uj = tan2 (45 +#2). Consequently for this special case of no volume change rate, where the stress ratio is identical to the ratio of energy input to energy output, in both triaxial and plane strain and where K is a maximum, the energy absorbed is a minimum with respect to all possible mean values of p for the chosen stress level. Also the absorbed energy has to be considered at a chosen stress or energy level. It has been shown (Rowe, 1964b) that if the ratio of incremental energy input in the major principal direction divided by the incremental energy output in the minor principal direction is E where U; de, EC- - uj de, in plane strain compression and u; de1 -_- E = -2ujdE3 E _ 27; dcl - u; de, in triaxial strain compression in triaxial extension Fig. 1. Coulomb minimum energy solution 78 I. W. ROWE then at any stage of a compression test the absolute increment of work absorbed dW is given by dW = crj delR 1-g [ 1 In a test one might choose the value of the cell pressure U; and keep it constant, and choose the stress level by selecting a value of R which may or may not be the peak, but which remains sensibly constant during a chosen small applied strain A,. The applied input energy level uj de,R is thus chosen. For this applied energy input and chosen stress level the absolute energy absorbed d W is a minimum when E is a minimum. One might call this an energy ratio principle but it does not invoke a new principle since only one of the four test variables in the above expressions is free to vary with E and the particle sliding direction; the other three are chosen and held constant. It is seen therefore that equation (1) does not invoke a new principle in respect of mini- mization of energy, but is merely a generalization of the Coulomb-Rankine expression to include dilatancy rates other than unity. It may also be noted that the flow rule of plasticity associated with the Mohr-Coulomb failure criterion constitutes a special case of the stress-dilatancy relation where the interparticle friction angle (b, =0 and in plane strain ~1 &Ju$ dr, = - 1. In an outstanding general treatment of rotund particles in contact, Horne (1965) not only verified equation (1) but extended the finding to show that for the case of triaxial compression the maximum possible value of D =2. Thus at peak 1~0~2 . . . . . . . . (7) It may be noted that equation (1) was derived originally with the aid of an equation R = tan CI tan (1$,+/3) . . . . . . . . (8) which applied to any formal packing or any two particles in contact. This equation was also derived independently by Biarez (1961), Leussink and Wittke (1963) and Parkin (1965). Insertion of the critical direction for a single pair of contacts and with allowance for deviations of individual contact directions from the mean critical direction led to the equation Horne derived R = tan cc 1<1/2 . . . . . . . . (9) R=;zKll . . . . . . . . 3 where ++Qn3 was the ratio of the number of particle contacts at incipient slip, in the principal planes. Rowe (1962) observed that tan a was a property of the particle contact arrangement as distinct from particle friction, and while this was supported by equation (10) Horne demon- strated that tan u was not independent of the value of 4,. In contrast the dilatancy rate D was independent of $, and of K. Experiments by Parikh (1967) using a variety of particle shapes, gradings and mineral composition of saturated sands are shown in Fig. 2(a). For this the initial relative porosity (Kolbuszewski, 1950) and that at peak was calculated on the assumption of uniform porosity throughout the sample. It is seen that at a given relative porosity, only extreme variations in particle shape, as between glass spheres and crushed glass, have a clearly defined influence on dilatancy. In general D varies approximately between the limits of 2 and 1 as deduced by Horne (1965). These limits have been derived only for triaxial compression and, although a similar tendency was noted by Khayatt (1967) in triaxial exten- sion and by Wightman (1967) in plane strain (Fig. 2 (b)), one might expect the range in plane strain to be somewhat larger. SHEAR STRENGTH OF SANDS 79 0 08 06 04 02 0 INITIAL RELATIVE POROSITY ZO I) t 0 0 I* n I5 m 0 . I.0 I.0 08 06 04 02 0 . A RELATIVE POROSITY AT PEAK (a) Fig. 2 (above). Experimental relations between dilatancy rate and relative porosity at low pressures Fig. 3 (right). Relation between &, and CCV I . 01 / / - - %. _ I.0 0. 8 0.6 o-4 0.2 0 5( $1 4[ 3[ 2c I( ( )- )- )_ )- I- ) INITIAL RELATIVE POROSITY @I IO 20 30 +; - 4c The experimental relation between 4, and +,, shown in Fig. 3, which is in agreement with a theoretical prediction by Horne, refers to a free mass of particles under water. The values of +, and d,, in Fig. 3 were determined from the slope of stress ratio-dilatancy rate plots of triaxial compression tests, on samples in dense states during pre-peak loading to give c$, and at large strains using free ends to give &. In the case of glass ballotini, quartz and feldspar it was possible to obtain very similar values of $, by means of a direct shear test of a mass of particles sliding over a polished block of the mineral under water, but no material in block form was available for the zircon or the crushed glass. PEAK STRENGTH LIMITS IN TRIAXIAL AND PLANE STRAIN COMPRESSIOS Tviaxial compression Taking the upper dilatancy rate D =2 and using equations (la) and (3) the upper limit for the stress ratio of a sand in the densest state is given by o~/ c$, = 2 tan2 (45+$,/ 2) . . . . . . . (11) Taking the lower dilatancy rate D = 1 and using equations (la) and (4) the lower limit for the stress ratio of a sand in the loosest state is given by CJ;/ IJ~= tan2 (45+&,/ 2) . . . . . . (12) 80 P. W. ROWE INITIALRELATIVE POROSIN Fig. 4. Relation between Coulomb &,,,, as measured in plane strain, triaxial compres- sion and direct shear 50 I (b) $,= 26 1 , J 0 05 IO INITIAL RELATIVE POROSITY - - Experiment v .I I .,T;;;i* , . P.S. I #T Glasr ballotinl o D.S. I b I 05 INITIAL RELATIVE POROSITY I I-O Plane strain compression Using equations (la) and (5) and assuming the same dilatancy limits in plane strain as for triaxial strain, namely dense $ = !2tan2 (45+&.,/ Z) *3 . . . . . . * (13) loose $ = tan2 (45+$,,/ 2) 03 . . . . . . - (14) These upper and lower limits are shown on Figs 4(a), (b) and (c) in terms of the Coulomb angle 4 max, using the Rankine equation, plotted at relative porosities of 1 and 0 respectively for each test type. The variation of dilatancy rate with relative porosity is not necessarily linear and is unknown, Experimental results show a relation between d,,, and relative porosity which is nearly linear and straight lines are drawn between the theoretical limits in Figs 4(a), (b) and (c). The limiting dilatancy rate of 2 is not necessarily reached by dense packings. Increase of mean principal stress, for example, leads to crushin, 5 and a reduction of dilatancy rate to an extent such that at very high pressures the maximum dilatancy rate is unity. However, for a given measured value of &,,X for one sand in one state and mean principal stress at failure in the triaxial test, the corresponding value of $max in plane strain may be deduced from a plot SHEAR STRENGTH OF SANDS 81 of the limiting values for the particular values of #, and &. For this purpose 4, may be determined during a reloading stress path in the triaxial test and Fig. 3 used to assess &, or the value of qGcy may be measured directly in the triaxial test using free ends and large strains. The value of & may differ slightly in plane strain from that in triaxial strain but more experimental evidence is required on this point. PEAK STRENGTH IN DIRECT SHEAR AND PL.4NE STRAIN COMPRESSION It has been noted (Rowe, 1958; Rowe, Barden and Lee, 1964) that the Coulomb 4 measured in direct shear must differ from that measured on a principal stress element in a compression test apparatus. This is in contrast to the result obtained when assuming the Coulomb failure criterion and the orientation of the principal stress element in a shear box such that the sliding plane coincides with the plane of applied shear as given for example by Taylor (1918). The problem was considered by Hansen (1961) who assumed coincidence between the direc- tions of principal stress and total strain. De Saint Venant (1870) inferred that for plastic deformations of isotropic materials the principal directions of stress and strain rate should coincide. Hill (1950), treating a perfectly plastic material, stated that if the principal axes of stress coincide with the axes of anisotropy, so do the principal axes of strain increment. A characteristic of the particulate model is that during a stress path under an increasing stress ratio a new sample particle structure is formed with a new set of contacts during each small stress ratio increment. The model is orientated in the applied direction of principal stress and if that direction changes during an increase in stress ratio, when interparticle slips predominate, the new orientation of critical contacts must be such as to achieve equilibrium with the new direction of applied stress. Two points arise. The frrst is that the orientation of the critical contacts controls the axes of anisotropy of the model, whose properties depend entirely on the contacts, so that there is conformity with the statement of Hill. Second, the coincidence of stress and strain rate direction in an element under test with fixed principal stress directions, prior to any formation of discontinuity or slip bands, ought therefore to apply equally to an element which undergoes a gradual reorientation of the directions of principal stress. Cole (1967) and Roscoe, Bassett and Cole (1967) have reported direct measurements of normal and shear stress on the boundary of an element of sand in the centre of the Cambridge S.S.A. Mk. 6 and assuming uniformity of strain throughout the element have shown, for the first time, coincidence of the directions of principal stress and principal strain rate during small incremental stress rotations. In their diagrams for Mohrs circles of stress and strain increment for the stress condition in simple shear they use the same angle # between the horizontal plane of the simple shear apparatus and the direction of principal stress on the one hand and between the horizontal plane and that of the principal strain increment on the other. A necessary consequence is that Mohrs circles of stress and strain rate are geometrically similar. This together with the condition of zero strain increment in the horizontal direction in direct shear (Fig. 5) leads to the following equations. From Mohrs circle of strain increments dv -= 2 (de, -d4 cos 2 + _ 2 . . . . . . . In two dimensional strain From equations (15) and (16) dv = dc,+dcg . . . . . . . . (16) cos 2# = dv dvldc, 1-D 2 de,-dv = i-dvldc, = - l+D 82 P. W. ROWE Axis of Axis of (a) Mohrr circle of sfress (b) Mohrr circle of strain incremenf Fig. 5. Mohrs circles of stress and strain increments From LMohrs circle of stress (T;+uj - =a-TCOt2* . . . . 2 and U;-U; n = 7 cosec2$ . . . . L From equations (18) and (19) u; + u; R+l -7--7=R_l cl-u3 = csin 2#--cos 2# [Ssing equation (17) R+l (T -=- R-l i- J &(l-D)2 (1-D) u 20 (I+0)2)=$D)) whence 7 I i- (R-l) ~=&~(R/ D+l) Inserting the flow rule equation (la) into equation (21) . 7 2= J I((R-1) R(K) ... . . where with reference to equation (5) K = tan2 (45++,,/ 2) . . . Substituting R = tan2 (45 +&,,/ 2) and T/ O =tan & equation (22) becomes tan & = tan & cos $,, . . . . . . . . (18) . . . w . . (20) . . (21) . . . (22) . . (23) . . (24) Cole (1967) obtained a numerical relation between T/ O and R and noted that it must depend on the value of +, and Roscoe, Bassett and Cole (1967) concluded that a relation between these parameters was imminent. has published the relation Also, since the original preparation of this Paper, Davis (1968) 7 cos 4 sin + 2 z-7 l-sm$sin+ between T/U in direct shear, 4 in plane strain and a parameter denoted here by z+8 which is defined by the identity D =tan2 (45+$/ 2). relation, equation (22) is obtained. Eliminating I$ and using the stress-dilatancy SHEAR STRENGTH OF SANDS 233 Equations (22) and (24) relate the peak stress ratio R or & in the plane strain test to the peak stress ratio T/ U or q& measured on a horizontal plane in simple shear, as a function of the critical state angle &, which in turn is a function of 4,. For the constant volume test as when R =I < and & =&, equation (22) simplifies to tan &s = sin q& . . . . (25) in agreement with Hill (1950) and as noted by Bishop (1954). Only in the imaginary case of frictionless particles with q5, =& =0 does & =&. The physical basis for this relation is that the orientation of the plane strain element within the shear box is governed by the restriction of movement such that the resultant horizontal strain increment is at all times zero. Consequently the critical slip plane in the plane strain element in general does not coincide with the horizontal plane of the box. The peak das applied to the horizontal plane must therefore be less than the &,, associated with the critical plane in the plane strain element which governs failure, and on the basis of the Coulomb theory this should lead to slip planes in the shear bos at angles inclined to the plane of the box. Some numerical values are given in Fig. 6 in which the difference (& -&) is plotted against +ds. One can insert any value of R from unity upwards and deduce the corresponding value of T/ O in equation (22), but the peak dilatancy rate range of 1 to 2 determines the range of R or Q-/ U for any given value of C,,. These are shown by solid lines in Fig. 6 for feldspar, quartz and glass sands, and it is interesting that for a given material the difference (+,, -&J is approximately constant over the entire dilatancy rate range. Working with the conventional shear box and its associated non-uniform strain pattern, Rowe (1954) observed values of & for Mersey River Sand between 23 and 42, while the range recently obtained by Wightman in plane strain compression over the same order of mean pressure range is 32-46 or some 4 higher, in close agreement with $,,=32 for that sand. It seemed therefore of interest to carry out conventional direct shear box tests on feldspar sand and glass ballotini at stress levels comparable with those used in plane strain. The results are shown in Fig. 6. These showed some 9 difference between shear box and plane strain values of dmax at the dense and loose limits for feldspar and some 2 difference for glass. Despite the objections to the conventional shear box it would seem that the experi- ments agree quite closely with theory over the wide range of q5, values investigated. Such Fig. 6. 4 *r Difference between &ax in plane strain and direct shear for various values of & 84 P. W. ROWE experiments need to be repeated in a direct shear apparatus which applies more uniform strain throughout the sample, but two points arise. First, if the theory is correct it would appear that the difference between peak values measured in a conventional shear box and those in an apparatus which allows much more uniform strain may be expected to be small. Second, there is still a fairly wide use of the direct shear test using various types of apparatus and the present results (Fig. 6) transferred to Fig. 4 indicate the relative magnitude of the Coulomb q5 to be expected from these tests and triaxial and plane strain compression tests. In this respect it may be noted that the trend for (&raxisl - $d,rect shear) to decrease and change sign passing from loose to dense states of a medium-fine quartz sand was noted by Nash (1953). Considering the direction of the Coulomb slip plane in the direct shear test, point A in Fig. 7(a) represents the stresses on the horizontal plane and AP is drawn horizontally through A to give the pole of the diagram at P. The failure plane of the principal stress element is then given by PB and the observed slip planes, if they formed at the peak strength, should lie at angle CL to the horizontal (Fig. 7 (b)). From the geometry of Fig. 7(a) I u3 sin & 7 sin B =(a;--oj)cosB . . . . . . and inserting p = 0 - & sin As cosOsin(O-$,,) = ~_I . . . . . . Substituting for tan q& (equation (24)) into equation (27) and rearranging e =COS-1 J-c ( 9 I-- sin & co? #J~ -sin +,, co? f& 1 - sin2 dDs sin2 q5,, II- . . . (27) (28) whence from Fig. 7(a) a =45+-p-e . . . . . . . . (29) For the special case of no volume change rate at failure, when & =&, equation (28) yields e=45 for all values of 4,. This is evident immediately from Fig. 7(a) since for this case Fig. 7. Slip plane direction in direct shear SHEAR STRENGTH OF SANDS 85 points A and P are both located at C. Consequently a=$& as deduced by Hansen (1961), but with the special value $,,/ 2. For sands dilating at the limiting rate D ~2, equation (28) yields 0=cos-11/ 1/ 3 =5+75, and using equation (29) a = %- k-75 - (~ +)I . . . . . , (30) Taking the values of 4, given in Table 1, and deducing &,, from the stress-dilatancy equation for D =2 it is seen that the value of a is about 2 less than q&/ 2 for quartz sands at the maxi- mum dilatancy rate. For intermediate dilatancy rates the value of a may be determined Table 1 0 19.5 0 10 28.7 0.40 20 37.3 I.10 30 45.6 1.8 40 53.4 3.0 from equations (28) and (29) using measured values of 4pS, measured & values with equation (24) or measured peak dilatancy rates with equation (1). In general for quartz sands It is clear that for dilating sands the value of a can be considerably smaller than +,,/ 2. This analysis should apply to drained tests on normally consolidated clays where c=O, but where a comparison is made between measured and calculated directions of slip surfaces, e.g. Morgenstern and Tchalenko (1967), it is strictly necessary to maintain reasonably uniform strain conditions with a Cambridge simple shear apparatus and to record the dilatancy rate at peak effective stress ratio. CONCLUDING DISCUSSION Many earth pressure and stability problems are those of plane strain, but owing to pro- gressive failure throughout the mass the 4 values necessary to fit earth pressures to simple failure theories for dense sand can vary over a wide range between limits close to the plane strain peak for an element &,ax in the active state and close to 4, in the passive state (Rowe, 1967). Average mass 4 values less than the plane strain peak are more appropriate and this may be one reason for the apparent success of the use of the triaxial test. The difference between the direct shear value and that from the triaxial test for quartz sands in the dense state is less than the uncertainty associated with use of a singular average Coulomb 4 value in the mass at failure, whereas in the loose state where movements are more serious the lower direct shear values are conservative. For sands, therefore, the direct shear test may still be attractive in practice. ACKNOWLEDGEMENT The Author is indebted to Professor M. R. Horne for permission to include his unpublished theoretical curve in Fig. 3, and to Professor A. W. Bishop and Professor K. H. Roscoe for improvements in presentation. 86 I. W. ROWE REFERENCES BARDEN, L. & KHAYATT. A. J. (1966). Incremental strain rate ratios and strength of sand in the triaxial test. Gdotechnique 16, No. 4, 338-357. BIAREZ, J. (1961). Contribution a letude des properties mecanique sols et des materiaux pulvCrulents. D.Sc. thesis, Grenoble University. BISHOP, A. W. (1954). Correspondence on Shear characteristics of a saturated silt measured in triaxial compression. G&technique 4, No. 1, 43-45. COLE E. R. L. (1967). Soils in the S.S.A. Ph.D. thesis, Cambridge University. DAVIS E. H. (1968). Theories of plasticity and the failure of soil masses. (ed. I. K. Lee). Butterworth. Soil mechanics, selected topics DE SAINT VENANT, B. (1870). Memoire snr letablissement des equations differentielles des mouvements interieurs opCrCs dans les corps solides ductiles an dclB des limites ou lelasticite pourrait les ramener a leur premier &tat. C. Y. hebd. Seanc. Acad. Sci., Imis 70, 473-480. EL-SOHBY, M. A. (1964). The behaviour of particulate material under stress. Ph.D. thesis, Manchester University. HANSEN, B. (1961). Shear box tests on sand. Proc. 5th Int. Conf. Soil Mech. 1, 127-131. HILL R. (1950). The mathenzatical theorv of plasticitv. HORNE, l&. R.(1965). Oxford Universitv Press. The behaviour of anassembly of rotund, rigid cohesionless particles. Parts I and II. Proc. R. Sot. A 286, 62-97. KHAYATT, A. J. (1967). Some incremental stress-strain relations for sand. Ph.D. thesis, Manchester University. KIRKPATRICK, W. M. (1961). Discussion on Les proprietes des sols et leur mesure. Proc. 5th Int. Conj. Soil Mech. 3, 131-133. KIRKPATRICK, \V. M. & BELSHAW, D. J. (1968). On the interpretation of the triaxial test. 18, No. 3, 336-350. Geotechnique KOLBUSZEWSKI, J. J. (1950). Notes on deposition of sands. Research, Lond. 3, 478-483. LEE, I. K. (1966). Stress-dilatancy performance of feldspar. J. Soil Mech. Fdns Dio. Am Sot. civ. Engrs 2, 79-103. LEUSSINK, H. & WITTKE, 1V. (1963). Difference in triaxial and plane strain shear strength. Symposium on laboratory shear testing of soils, 77-89. American Society of Testing and Materials, S.T. Pub. 361. MORGENSTERN, N. R. & TCHALENKO, J. S. (1967). Microscopic structures in kaolin subjected to direct shear. GCotechnique 17, So. 3, 309-328. NASH, K. L. (1953). The shearing resistance of a fine closely graded sand. Proc. 3rd Int. Conf. Soil Mech. 1, 160-164. PARIKH. P. V. (1967). The shearing behaviour of sand under axisymmetric loading. Ph.D. thesis, Man- Chester University. PARKIN. A. K. (1965). The application of discrete unit models to studies of the shear strength of granular materials. Ph.D. thesis, -University of Melbourne. - _ PROCTER, D. C. (1967). The stress-dilatancy behaviour of dense sand in the hollow cylinder test. M.Sc. thesis, Manchester University. ROSCOE. K. H. (1967). Discussion on Shear strength of soil other than clay. Proc. geotechnical Co~f. 2, 188-192. ROSCOE, K. H., BASSETT, R. H. & COLE, E. R. L. (1967). Principal axes observed during simple shear of a sand. Pvoc. geotechnical Conj. 1, 231-237. ROWE, P. W. (1954). A stress-strain theory for cohesionless soil with applications to earth pressure at rest and moving walls. Gtotechnique 4, No. 2, 70-88. ROWE, P. W. (1958). Closing remarks session 1, Proc. Conf. Earth Pressures. ROWE, P. W. (1962). The stress-dilatancy relation for static equilibrium of an assembly of particles in con- tact. Proc. R. Sot. A 269, 500-527. ROWE, P. W. (1963). Stress-dilatancy, earth pressures and slopes. Proc. Am. Sot. civ. Eplgrs 89, 37-61. ROWE, P. W. (1964a). Discussion on Some experiments on the influence of strain conditions on the strength of sand. Gdotechnique 14, No. 4, 361-364. ROWE, P. W. (1964b). Closure to discussion on Stress-dilatancy, earth pressures and slopes. Proc. Am. Sot. civ. Engrs 90, 145-180. ROWE, P. W. (1967). Discussion, Session 3. Proc. Geotechnical Conf. 2, 210-211. ROWE, P. W., BARDEN, L. & LEE, I. K. (1964). Energy components during the triaxial cell and direct shear tests. Gdotechnique 14, No. 3, 247-261. TAYLOR, D. W. (1948). Fundamentals of soil mechanics. Wiley. WIGHTMAN, A. (1967). The stress-dilatancy of sands during plane strain compression. M.Sc. thesis, Manchester University.