You are on page 1of 58

BISHOP, A. W., GREEN, G. E., GARGA, V. K., ANDRESEN, A. & BROWN, J. D. (1971).

Gtotechnique 21, No. 4, 273-328.

A NEW RING SHEAR APPARATUS AND ITSAPPLICATION


TO
THE MEASUREMENT OF RESIDUALSTRENGTH
A. W. BISHOP,* G. E. GREEN,? V. K. GARGA,$ A. ANDRESEN/~ and J. D. BROWNS
SYNOPSIS
This Paper describes both the design and principles Cet article decrit a la fois la conception et les
of operation of a new ring shear apparatus, and its principes d’operation d’un nouvel appareil de
mesure du cisaillement circulaire par torsion, et de
application to the measurement of the residual son application a la mesure des resistances residuelles
strength of undisturbed and remoulded samples. d’echantillons non pertubes et remoules.
As the test results differ significantly from those Comme les resultats des essais different d’une
obtained in multiple reversal direct shear box tests, man&e significative de ceux obtenus lors d’essais
multiples effect&s dans une boite de cisaillement
a critical examination is made of all possible sources renverse direct, on donne une analyse critique de
of error in both measurement and interpretation. toutes les sources d’erreurs possibles 1 la fois dans
The results of tests on five soils, blue London Clay, les mesures et dans leur interpretation.
brown London Clay, Weald Clay, a Norwegian clay Les resultats d’essais sur cina ~01s: arsdle bleu de
Londres, argile brun de Londrks, argile”de Weald,
and the Cucaracha Shale from Panama, are presented argile Norvegien et schiste Cucaracha de Panama,
and discussed in relation to other published residual sont present% et discutes par rapport aux autres
strength data from tests on these materials. The resultats d’essais de resistance rcsiduelle publies sur
ces materiaux. On examine les facteurs qui con-
factors controlling the brittleness of soils tested under trolent la fragilite de ces sols dans des conditions de
drained conditions are examined. drainage.

INTRODUCTION
In his paper ‘Torsion shear tests and their place in the determination of the shearing re-
sistance of soils’, M. Juul Hvorslev (1939) stated that the principal objects of shear tests were
the determination of:
(a) the maximum shearing resistance,
(b) the bond resistance and the velocity of the slow plastic flow before failure,l
(c) the temporary or permanent decrease of the shearing resistance after failure, and
(d) the stress-strain relationships and volume change characteristics due to shearing
stresses.
In addition Hvorslev emphasized that consideration had to be given to simplicity in the
construction and operation of the testing apparatus, to difficulties in sample preparation, and
to the time required for testing. After examining the various tests available, including the
direct shear test using the shear box, he concluded that objects a, b and d could best be
achieved in the triaxial apparatus.
With regard to object c, Hvorslev added that ‘the fact that very large displacements often
are required to produce this minimum value of shearing resistance seriously affects the suitability
of various types of shearing apparatus. The data on the decrease of the shearing resistance
after failure are required to determine the factor of safety of earth structures in which a localized
failure of the soil is allowed or cannot be prevented.’ Hvorslev concluded that this object
could be achieved satisfactorily only in a form of the torsion shear test which he described as
the ‘ring shearing apparatus’. Various forms of this apparatus had been developed indepen-
dently in about 1934 by himself, by Gruner and Haefeli, and by Cooling and Smith.
* Professor of Soil Mechanics, Imperial College, London, England.
t Lecturer in Civil Engineering, Imperial College, London, England.
: Formerly Research Student at Imperial College, now Senior Engineer, Hidroservice, Sao Paulo, Brazil.
// Head of Mechanical Design Section, Norwegian Geotechnical Institute, Oslo, Norway.
§ Formerly Research Engineer, Norwegian Geotechnical Institute, now Assistant Professor, Nova Scotia
Technical College, Canada.
r The study of creep characteristics.
273
274 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

SHAPE OF
SAMPLE
FAILURE LOADlNt SYSTEM REFERENCE
TYPE
SURFACE

SOLID CYLINDER
LOADED NORMALLY,
A.S.C.E. (1917)
CIRCLE LOWER PLATEN
TWlSTED

SOLID CYLINDER STRECK (1928)


CIRCLE LOADED NORMALLY FRANZILlS ET AL.(l929)
‘ TWISTED (se. HVDRSLE”, 1939)

SOLID CYLINDER
LANCER (1938)
CIRCLE LOADED NORMALLY
(se. HVORSLEV, 1939)
‘ TWISTED

SOL,0 OR DlVlDED CONFINING


I ( RINGS
oii \IdT
SOLID DISC SEMBENELLI L

CIRCLE LOADED NORMALLY RAMlREZ (l9b8)

L TWISTED LA CATTA ,970

SOLID DISC
TIEDEMANN(l937)
ANNULUS LOADED NORMALLY,
(a HVORSLEV, 1919)
ANNULUS TWISTED
ROTATE L HEASURE r

0ii t 7. MEASURED

SOLID DISC

CIRCLE LOADED NORMALLY GHANI (1966)

L TWISTED

CASACRANDE L
HOLLOW CYLINDER LLS.ENtlNEER OFFICE,
CYLINDER. LOADE D RADIALLY BOSTON, MASS.
L TWlSTED (we HVORSLEV, 1939)

Fig. 1. Principal features of various forms of torsion and ring shear apparatus:
(a) solid cylinder or disc or hollow cylinder
A NEW RING SHEAR APPARATUS 275

SHAPE OF
SIMPLE
F*IL”RE LOADING SYSTEM REFERENCE
TYPE
SURFACE

UNCONFlNED ANNULAR
COOLING C
ANNULUS DISC LOADED
SHITH (1915,l936)
NORMALLY C

TWISTED

CRUNER t
ANNULAR DISC
HAEFELI (1934)
ANNULUS LOADED NORMALLY
HAEFEL, (19JI)
C TWISTED
(8,. HVORSLEV, 1939)

ANNULAR DISC TlEDEMANN (1937)

LOADED NORHALLY (me HVORSLEV, 1919)


t TWlSYED

ANNULAR DISC HVORSLEV(l937,1939)

LOADED NORMALLY HVORSLEV ‘


ANNULUS
KAUFMAN. (1952)
t TWlSTED
HERRMANNC WOLFSKILL
(1966)

SOLID OR DIVIDED OUTER

ANNULAR DISC

ANNULUS LOADED NORMALLY LA CATTA (1970)


C TWISTED
l---km 4

ANNULAR DISC

ANNULUS +@1;fi LOADED NORMALLY NOVOSAD (l9b4)


C TWISTED
bcm
,I<”

ANNULAR DISC
CARR L
ANNULUS LOADED NORMALLY
WALKER (1968)
‘ TWISTED

ANNULAR DISC
SCARLETT L
ANNULUS LOADED NORMALLY
TODD (1968)
‘ TWISTED

Fig 1. Principal features of various forms of torsion and ring shear apparatus :
(b) annular disc
276 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

Subsequent events have shown Hvorslev to be correct. The major part of all subsequent
research work on the peak strength of soils and on the relationships between stress, strain,
volume change and time has been carried out in the triaxial apparatus.2 Renewed interest in
the post-peak behaviour of clay soils stimulated by Skempton’s 1964 Rankine Lecture has led
to a re-examination of the problem of measuring the shearing resistance at very large displace-
ments. The results presented in this Paper confirm that in relation to object c, Hvorslev’s
conclusions also hold good.
The lack of use made of the ring shear apparatus in the intervening years may at first sight
seem surprising. However, it certainly failed (and still fails) to satisfy the criteria of simplicity
of construction and operation, and the duration of the test tends to be long, unless relatively
thin specimens are used. Moreover, for purposes other than the measurement of residual
strength it has all the disadvantages of the shear box, such as high local concentrations of strain
and uncertainty about the directions of the principal stresses as the test proceeds.
One of the principal mechanical difficulties in carrying out a ring shear test is that of main-
taining a small but adequate clearance between the two pairs of metal rings enclosing the upper
and lower sections of the annular test specimen, in order to minimize metal to metal friction or
loss of soil during shear.3
A novel method of overcoming this difficulty was proposed by the senior Author and formed
the basis of the design project, which was carried out jointly by the Norwegian Geotechnical
Institute and Imperial College.4
Before the new apparatus is discussed in detail, a brief review of earlier work on other forms
of ring shear apparatus is given.

REVIEW OF EARLIER WORK ON TORSION AND RING SHEAR APPARATUS


In general, the two obvious advantages of any type of torsion or ring shear test are that
there is no change in the area of the cross section of the shear plane as the test proceeds and
that the sample can be sheared through an uninterrupted displacement of any magnitude.
The latter, in particular, renders it very suitable for determining the residual strength of clays.
The problem is to design an apparatus in which the normal and shear stresses on the failure
plane are as near-uniform as possible and the consequences of their not being uniform are
known and of acceptable magnitude. A convenient method of sample preparation is also
required and the apparatus must be capable of transmitting to the sample the desired combina-
tions of normal and shear stress.
The simplest test of this type is one in which a laterally confined normally loaded solid disc
or cylinder is twisted, Fig. 1 (a). The device used by the American Society of Civil Engineers
(1917) and the similar apparatus of Streck (1928), F ranzius et al. (1929), Langer (1938),
Tiedemann (1937) and Ghani (1966) all suffer from the inherent defect that failure must occur
adjacent to the rotating normal load platen and not within the body of the sample. In the
case of both the American Society of Civil Engineers’ and Streck’s apparatuses the radial dis-
tribution of stress and strain is likely to be very non-uniform. In both Tiedemann’s and
Ghani’s apparatus the divided annular platen over which the shear stress is determined offers
an improvement in this direction. However, should differential vertical movements occur

1 With the exception of work at Cambridge on the critical state for which the direct (or simple) shear test
has been used. This apparatus in its original form has been described by Roscoe (1953). In this connexion
it is of importance to draw attention to the difference between the critical state in cohesionless partldate
materials having relatively rotund particles and the residual state in clay soils consisting of platey particles
subject to strong inter-particle forces. In the former case, yield at approximately constant void ratio and
constant effective stress can be achieved at relatively small strains (within the capacity of conventional testing
equipment). In the latter case the localized high degree of particle orientation associated with the minimum
value of shear strength is achieved only after much larger deformations.
3 A similar difficulty arises in the conventional shear box.
’ The new ring shear apparatus was manufactured by Geonor A.S., Oslo, Norway.
A NEW RING SHEAR APPARATUS 277
between sections of the platen, arching will develop and the measured stresses will be incorrect.
The divided confining rings used by Langer (1933) and, more recently, by Sembenelli and
Ramirez (1969) and by La Gatta (1970) are a marked improvement as they tend to promote
failure within the body of the sample at mid-height, remote from the normal loading platens.
However, the extreme non-uniformity of strain and possibly of stress remains. This difficulty
can be minimized by using a ring-shaped sample of suitable dimensions so that a less non-
uniform distribution of displacement and shear stress exists across the radial width of the
sample.
The earliest ring shear apparatus for testing soil appears to be that designed by A. Casa-
grande in collaboration with the U.S. Corps of Engineers, Boston (Hvorslev, 1939). A hollow
cylinder, confined by end platens, is loaded radially and rotated so that failure occurs on a
cylindrical surface, ideally at mid-wall thickness (Fig. 1 (a)). Sample preparation, shear load
transmission and mechanical difficulties appear to be considerable.
The most promising type of loading system as applied in its various forms to a ring-shaped
specimen is shown in Fig. l(b). Cooling and Smith (1935, 1936) used an unconfined annular
disc, loaded normally and twisted. Since the soil is unconfined, side-friction problems are
eliminated but the normal load that can be applied is very limited and shear stress transmission
appears to be problematic. Divided confining rings in varying forms were introduced by
Gruner and Haefeli (1934), Haefeli (1938), Tiedemann (1937), Hvorslev (1937, 1939, 1960),
Hvorslev and Kaufman (1952) and more recently by La Gatta (1970). The position of the
division in the confining rings in Haefeli’s apparatus appears to be ill-chosen since it encour-
ages failure near the lower platen. This difficulty is avoided in Tiedemann’s and Hvorslev’s
apparatuses, which are very similar. The proportions of the samples used in the latter case
are such that radial stress variations are minimized while sample preparation remains prac-
ticable and the influence of friction between the sample and the confining rings and the stress
concentrations at the edge of the raised rings seems likely to be small. A most important
feature of Hvorslev’s apparatus is the provision of divided confining rings so that, on the one
hand, no portion of the normal load is transmitted across the confining rings and, on the other,
that the gap between them is small enough to prevent loss of soil by squeezing.
Renewed interest by A. Casagrande has led to the development at Harvard of a rotary
shear apparatus capable of testing either a solid disc or an annulus, but of varying height, O-l-
2.5 cm (Sembenelli and Ramirez, 1969; La Gatta, 1970). Generally a very thin remoulded
sample, 0.1-0.3 cm thick, is retained by PTFE confining rings so that side-friction effects are
small. However, if the divided outer rings are used, excessive squeeze affects the normal
stress distribution and the advantages of the thin sample are lost. Tests by La Gatta (1970)
on blue London Clay from Wraysbury (referred to in detail later) using an annular sample
show identical residual strengths to those measured with the ring shear apparatus described in
this Paper. This agreement is a very strong indication that both apparatuses (using annular
samples) are capable of accurate residual strength measurements.
Recently, attempts have been made by chemical engineers to design a rotary shear appara-
tus to measure the shear properties of dry powders. That used by Novosad (1964) is similar
in principle to Hvorslev’s apparatus whereas those of Carr and Walker (1968), and Scarlett
and Todd (1963) are of a less refined design and suffer from many of the inadequacies dis-
cussed earlier.
A fuller discussion of many of the principles involved in the ring shear apparatus is given by
Hvorslev (1939).

PRINCIPLES OF NEW APPARATUS

An annular ring-shaped specimen (Fig. 2), subjected to a constant normal stress u’,,, is con-
fined laterally, and ultimately caused to rupture on a plane of relative rotary motion. The
278 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

I
Axis

Plane of relative
rotary motion

Fig. 2. Ring shear test sample

new apparatus (Fig. 3) is designed so that the total normal load and shear torque being trans-
ferred through the soil across the plane of relative rotary motion are accurately known, i.e. fric-
tion forces in the apparatus are demonstrably minimized or are measured where appropriate.
The inner diameter of the sample is sufficiently large relative to the outer diameter for uncer-
tainties arising from an assumed non-uniform stress distribution across the plane of relative
rotary motion to be reduced to an acceptable level.
The sample, confined between pairs of upper and lower confining rings, is loaded normally
through annular loading platens by a dead-load lever system. The normal load is trans-
mitted by a vertical main shaft mounted in ball bushings to accommodate both linear and
rotary motion. The lower half of the sample is carried on a rotating table driven by a worm
gear. The upper half of the sample reacts via a torque arm against a pair of fixed proving
rings that measure the tangential (shear) load. The gap between the upper and lower con-
fining rings can be controlled and the side friction can be measured by means of a guided
linking yoke and a proving ring connected by a screw to the rigid crosshead. Hence the gap
between the rings is controlled relative to a fixed datum, and provided the proving ring for
measuring side friction is stiff enough, this gap will require little or no adjustment as the side-
friction force changes owing to the tendency of the sample to dilate or contract as the test
proceeds.5
The new ring shear apparatus is a controlled rate of displacement device since the primary
purpose of the tests is to study the post-peak section of the stress-displacement curve. The
gross normal load is maintained constant during shear and the net normal load is obtained at
any stage of the test by subtracting (or adding) the measured side-friction force. The varia-
tion in the net normal load during shear is very small and can be allowed for in the analysis
with no loss of accuracy, since the object of the test is to define the ratio of 7 to u’,, under
drained conditions. Maintenance of a constant net load is neither easy nor necessary.6

PRINCIPAL FEATURES OF NEW DESIGN


The general layout of the new apparatus is shown in Figs 4-7 together with details of the
sample assembly in Fig. 8. The sample, o.d. =6*0 in., i.d. =4*0 in., initial thickness=0*75 in.,

5 This is in contrast to the system used by Hvorslev (1939)) Hvorslev and Kaufman (1952) in which the
upper confining rings are located relative to the upper loading platen and will require more frequent adjust-
ment, if, on the one hand, clay is not to be lost as the sample dilates or if metal to metal friction is to be
avoided as the sample consolidates during shear. Such adjustments will of necessity disturb the sample
more frequently than would appear desirable.
6 The tests on Studenterlunden clay performed at N.G.I. were carried out at constant net load by adding
additional weights to the hanger to compensate for side friction.
Boll bushings to occommodo Screw for controlling qop
linear and rotary motion between confininq rings

Rigid crosshead

Provinq ring for measuring


toadinq yoke and torque arm sjde frictipn on confining
rrnqs
Torque-free swivel

Linking yoke for upper


Provinq ring fo confining rings
tangential load Main shaft

Loadinq platen
Water both

Confininq rings

Rotating table

x1\\\\\\\
Aliqnment and thrust Bose

Adjustable support i,/ i


’\ 3
t
*Dead load

’ -
\‘-‘I
\ ~5
and qeorbox
Motor

\\W;// N/IA,,<

Fig. 3.
r----l
Principles of operation of Imperial College-Norwegian Geotechnical Institute ring shear apparatus
(simplified section)
280 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

can be subjected to a maximum nominal normal stress of 142 lb/sq. in. and a maximum nom-
inal shear stress of 71 lb/sq. in.
The cast aluminium alloy base carries two vertical rigid steel columns which provide reac-
tions for the torque arm and also support the crosshead which carries the upper confining ring
lifting mechanism. The worm gear that rotates the sample assembly is mounted in a
housing on the base. The worm gear is chain-driven by an electric motor and variable speed
gear box unit, which is fixed to the floor beneath the apparatus to minimize vibration effects.
The lever loading system controlling the normal load is mounted beneath the sample within
the base.
The apparatus can be described in detail by subdividing it into four components :
(u) sample assembly,
(b) normal loading system,
(c) confining ring gap control mechanism, and
(d) torque measuring system.

Sample assembly
The annular sample is laterally confined between two pans of rings (Fig. 8), and is loaded
normally (vertically) through annular platens. Drainage is obtained by means of porous
ceramic annuli (A80 KV vitrified bauxilite) screwed to the platens. In order to minimize the
risk of slip occurring at the soil/ceramic interfaces, 12 sharpened radial beryllium copper fins,
0.010 in. thick, projecting 0.080 in. and extending the full width of the sample, were provided
on the exposed face of each ceramic annulus. The lower confining rings and the loading
platen are screwed to the baseplate (Fig. 6). An alignment ring locates the outer confining
rings during the pre-shear stages but is retracted just before shearing begins (Figs 6 and 7).
The upper and lower confining rings are shown in Fig. 8, and are held together by locking
screws, which are removed before shear commences.
A Perspex water bath serves to prevent the sample from drying out during testing. The
confining rings and platens are of brass, plated to minimize corrosion effects in the presence of
an aggressive pore fluid.

Normal loading system


The vertical normal load on the sample is maintained by dead weights applied via a 10: 1
ratio lever arm (Fig. 5). The lever’s fulcrum on the base of the apparatus is adjustable ver-
tically (by 0.5 in.) so that the lever can be levelled to accommodate changes in sample thick-
ness. An adjustable counterweight is provided to initially balance the lever, main shaft and
torque arm before loading the sample. The normal load is carried by the main shaft and
transferred to the torque arm by a nut incorporating a spherical seating. The latter allows the
torque arm to tilt by a limited amount if the sample thickness varies differentially during a
test. The load is finally transferred to the upper annular loading platen by two curved seg-
mental spacer blocks screwed to the torque arm. Four guide pins align these components
(Fig. 8) and transmit the shear load. At the lower end of the main shaft a torque-free swivel

r Whether to use this type of thin vertical blade instead of a series of triangular wedge-shaped teeth as
suggested by Hvorslev and Kaufman (1952) was controversial. A series of thin vertical blades are easily
pushed into London Clay with limited local disturbance. Load is transmitted into the sample through a
series of small passive wedges and there is evidence (Fig. 14(b)) that an inclined slip surface may form
emanating from the tip of a blade. The effect of these on the measured value of qr is uncertain but seems
likely to be small. Triangular teeth of the form suggested would likely be difficult to embed into clay, par-
ticularly under low normal loads, and would disturb the whole of the sample to the full depth of the teeth at
least. The problem is most acute in the case of a quick clay or a clay shale where neither of these two
methods may be satisfactory. The N.G.I. has used a pm-cushion of needles or a layer of water-absorbing
gypsum plaster, whereas Hvorslev and Kaufman (1952) refer to the use of a gelatine layer in Sweden. The
apparatus may be readily modified to suit whichever technique is found to be most suitable for a particular
Proving ring and dial gouge

Differential screw

Confining ring 90 torque-free housing

Sample thickness dial gauge Sample thickness dial qau9e

Dial gauge bridge

anqentiol load proving rinq

Vernier for anqular displacement


Tangential load provinq rinq
Sample assembly

Worm qear housinq


Drive to warm

Levellinq *crew fo Torque-free swivel


lever loadinq arm

- Lever loadinq arm

Fig. 5. General layout of ring shear apparatus


Ball bearing to accommodate - Main shaft
linear and rotary motion
1 ,pSpherical seatinq
Proving rinq for
measurinq tanqeniial ‘load 0-rinq seating
-
I ‘I
Torque arm ! -E
- =
SpacersO)

Dowel pins (2) -


(for tranrmittinq torque) -
--
Baseplate
Rigid columns (2) c
Rotatinq table

Worm qear housing


Radial deep-groove -
boll bearinqs -
--
- -

Thrust. ball bearinqs


with spherical seatings

Fig. 6. Longitudinal section showing loading yoke and torque arm


Linkinq yoke for uppe j- Fastening screws (4)
confining rinqs

Guide pins (2)

Ball bearings to accommodate


linear and rotary motion Fasteninq screws (41

r
/------Angular displacement

Perspex water bath

J- rina seals

Oil seal

=Oil seal

Handle

Loading shackle .r Torque-free iwive!

Fig. 7. Transverse section showing linking yoke for upper confIning rings
284 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

Lockinq screws (3)

Perrpex water bath

Annular porous ceramic (2)


Aliqnment rinq for
pre-shear staqes

Lower annular
loadinq platen confining ring

Fig. 8. Details of sample assembly. Sample dimensions: o.d.=6*0 in.; i.d.=4*0 in.; initial
thickness =O-76 in.

permits the shaft to rotate slightly as the tangential load proving rings deflect and also pre-
vents undesirable bending stresses developing in the lever arm/main shaft connexion. A
handle is fitted to this torque-free swivel to permit it to be oscillated during shear to check its
freedom and break any static friction that may have built up.

Con$ning ying gap control mechanism


A linking yoke (Figs 5 and 7) is mounted over the sample on the main shaft with a ball
bushing to allow both linear and rotary motion. Four pairs of screws connect the upper con-
fining rings to this yoke so that the rings can be raised or lowered by a differential screw on the
crosshead. Two guide pins (Fig. 7) align the linking yoke with the upper loading platen to
ensure that during shear the upper confining rings and the top loading platen rotate as a unit,
while differential vertical movement is possible. The side friction is measured on a stiff
proving ring which operates either in compression or in extension. A torque-free housing be-
tween this proving ring and the linking yoke permits the latter to rotate slightly as the tan-
gential load proving rings deflect. A bridge supports two pairs of diametrically opposed dial
gauges measuring vertical movements; one pair for the upper loading platen, and the second
pair for the upper confining rings.

Torque measuring system


The sample is sheared by steadily rotating the lower half while the upper half reacts against
the torque arm. The worm-driven rotating table, on which the sample assembly baseplate is
clamped, is aligned axially on two radial deep-groove ball bearings. The shear torque is deter-
mined from the readings on two opposed tangential load proving rings (Fig. 5) mounted on the
rigid columns. Each proving ring carries a vertically aligned wheel, which can roll on a
hardened steel plate set in the torque arm (not shown) and which transmits an axial load to the
ring while allowing the torque arm to move vertically as the sample dilates or consolidates.

SAMPLE PREPARATION

Tests were carried out on two types of sample : remoulded and undisturbed.
The preparation of a sample of remoulded clay presents no difficulty. The confining rings
A NEW RING SHEAR APPARATUS 285
and the lower annular loading platen are assembled on the baseplate8 as shown in Fig. 8, the
lower ceramic annulus having previously been de-aired in boiling water. Wet clay in 1 in.
wide, $ in. thick strips is kneaded in, taking care to avoid entrapped air. A scraping tool
located on the upper confining rings is used to clean off the clay to a uniform 0.75 in.
thickness.
The undisturbed samples were brought from the field in tins either as hand-cut block
samples, 9 in. x 9 in. x 12 in., sealed in aluminium foil and paraffin wax, or as hand-cut 12 in.
dia. x 24 in. high cylinders enclosed in a rubber membrane. The steps in preparing an
undisturbed ring shear sample are shown in Fig. 9. Considerable difficulties in obtaining
satisfactory undisturbed ring shear samples of stiff fissured clays were initially anticipated.
However, experience so far at I.C. indicates that the core cutter method of preparation is
very satisfactory for over-consolidated clays of low sensitivity and medium stiffness, which are
free from stones. It may require some modification for sensitive and quick clays, or for
clay shales.
A thin-walled core cutter is pushed into a block, Fig. 8,stage (a), and separated from it with a
wire saw (b). The stiffening ring is removed and the upper clay surface is planed with a
straight-edge (c). The sample is inverted and located on the assembled outer confining rings
(d). A wooden piston9 mounted in a drill press is used to push the sample into the confining
rings to locate against a wooden spacer (e) after which the exposed clay surface is planed to its
final 0.75 in. thickness, flush with the upper confining ring (f). The 1.7 in. thick spacer is re-
placed by a thinner (nominally 0.9 in.) two-piece spacer, the inner disc section of which is
slightly shallower than the outer annular section (g).‘” The sample is pushed further into the
rings to contact the annular spacer (h) . The sample assembly is rested and aligned on the drill
table and a thin-walled core cutter pushed l1 firmly into the sample (i) and Fig. 10. The lower
annular loading platen and ceramic annulus together with the inner lower confining ring are
screwed to the baseplate (i). The outer confining rings with the sample are then added (k).
At this stage the underside of the sample is just in contact with the 12 radial fins. The sample
is embedded on the fins (1) and the upper inner confining ring is carefully inserted by hand (m)
and clamped to the lower inner ring (Fig. 8).
The sample assembly containing either a remoulded sample or an undisturbed sample is
screwed to the rotating table and the water bath fitted. The upper loading platen is aligned
and gently placed in position on the sample, resting on the fins. The torque arm is lowered
over the main shaft, the nut and spherical seating fitted and the lever arm counterweight
adjusted to achieve a balance. The sample assembly is aligned with the torque arm so that
the four guide pins (Fig. 8) can be inserted, the fins pressed into the sample and the main shaft
nut tightened. After the proving ring has been zeroed to balance the weight of the linking
yoke, the crosshead carrying the linking yoke is mounted on the rigid columns. The two guide
pins and eight fastening screws are inserted on the yoke (Fig. 7). The dial gauge bridge is
assembled, the lever loading arm is levelled and initial readings of the four dial gauges noted.
Appropriate dead loads are added to the hanger to consolidate the sample and the water
bath is filled. The angular displacement scale and vernier are mounted to complete the
assembly of the apparatus. After consolidation and before shearing, the various locking
screws are removed and the alignment ring is retracted (cf. Fig. 8 and Figs 6 and 7).
On completion of the shear stage of the test, which is described in the next Section, water is
siphoned out of the water bath and the dead load removed from the hanger. The angular dis-
placement scale, dial gauge bridge, crosshead and linking yoke and torque arm are successively
8 Index marks on all components ensure their correct rotational alignment.
9 The piston and subsequently described wooden spacers were faced with Formica where in contact with
the soil.
lo It would have been better if both sections of this spacer had been of identical thickness with a small
annular groove in the edge of the inner section to accommodate the core cutter (i).
11 The drill chuck is, of course, not rotated.
f
I

Wooden block

+in. thick IWO-


piccc spacer.

(9) P$aa I.7 in. spacer by 0.9 in. two-piece


(4 z;E;;d core cutter pushed into clay block
(h) Push clay further into rings to rest against two-
(b) Slice through block with wire saw
pece spacer

i.d.=&bin (~Szmm) . Q06Sin.


(I*bSmm) ; Thin-welled
I I- core cutter

,,~~,,

(c) Trim off of clay


uppersurface until flush with
core cutter

piecr spacer
(i) Remove inner core with cutter
Drill press

Piston
t-

Oufar confining

- I.7in. thick rpac~ (j) Pre-assemble lower annular loading platen with
porous ceramic and inner lower confining ring
(d) Assemble core cutter on outer confining ring on baseplate
(k) Place outer confining ring assembly and sample
(e) Zhn$ into rings to re*t against spacer (using on baseplate
drill press as guide) (I) Embed clay on to radial fins on lower porous
(f) Remove core cutter and trim off upper surface ceramic
of clay until flush with confining ring (m) Insert upper inner confining ring (see Fig. 8)

Fig. 9. Preparation of an undisturbed ring shear sample of London Clay


+ 4C

+ 3.c

6ILinear I
+ 2.c
-4
%
L
z +1x
2
b
‘-
-5

?
E C
o
L
f
;;
$$
- 1.c

- 2.0

_ TParabolic II

a
I

0 5
@uniform deqreer

Fig. 11. Influence of assumed stress distribution on value of angle of shearing resistance 9 determined from torque and normal load
A NEW RING SHEAR APPARATUS 289

Table 1. Equations for angles of shearing resistance obtained from drained ring shear tests
assuming different normal stress distributions

Normal stress P
non-uniform tan 4’ tan 4’ rl
distribution Puniform tan +’ for r,-2.oin ,a,;;
uniform
r,.3*0in

0.395 -f 1.000

P(r,-r) 2M (a + 2rl
2.14 0.424 + I.075
.O- 7) W.‘(r:+2r, r,+3r,‘)

P -P(r-r,) I.67 0:943


(a-r,)

P (r* - r)* I.116


3.33
(rl-rl)l

P(r,- r)’ 5M r, + 3r,


Parabolic II 2.73 0.362 ! 0.916
P w zi’ (6r: +3r, rl + rfj w

; Parabo’ic -m: 4P(r;;)($r) I.50 y - (3r:‘:;;;;+3r:) 0.397 $ I.006

Parabolic Ip P(2r - r, - rJ1 5M (r, + rl


3.00 - . 0.990
(rz - rJ 2w (2rlz+ r, r, +2rf)
P
290 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

removed. The problem now is to remove the sheared clay sample from the confining rings
with minimum disturbance so that it may be inspected and specimens taken for water-content
determinations, thin sections for fabric analysis, etcetera. The sample assembly and base-
plate are unfastened, removed from the rotating table, inverted and placed on a 3 in. high
metal ring (o.d. = 58 in., i.d. =4*4 in.) with the upper annular loading platen resting on this
ring. The sample assembly is unscrewed from the baseplate and the latter removed. The
assembly is then placed the right way up on the 3 in. high ring. The lower outer and inner
confining rings are removed by gently pushing downwards and the upper pair by sliding up-
wards while applying gentle thumb pressure to the upper loading platen. Extreme care must
now be exercised to remove the upper loading platens without disturbing the sample. A thin
spatula is inserted at the clay/ceramic interface, working around the sample between the fins.
The exposed sample may then be dissected as required.

TEST PROCEDURE
After allowing the sample to consolidate (or swell) under a constant normal load on the
hanger the alignment ring is retracted and the locking screws connecting the confining rings
are removed. The gap between the upper and lower pairs of confining rings is opened (initially
by 0~001-0~002in.) by means of the differential screw. A speed is selected to give a constant
rate of strain, that permits adequate dissipation of excess pore-water pressure, the clutch is
engaged and shearing commenced.
The speed was sometimes increased by 5-25 times the initial slow speed after peak strength,
to obtain large deformations more quickly, the gap usually being closed at this stage. The
shearing rate was then reduced to the original value, the gap reopened and readings were taken
over a period of time to confirm whether or not a steady state had been achieved. This speed-
up/slow-down process could be repeated, so as to define a stress-displacement curve whose
range was limited only by the time available for the test.

CALIBRATIONS AND EVALUATION OF THE RESULTS


The two tangential load proving rings, each of capacity 100 lb, were removed from the
apparatus and calibrated in compression by dead weights. The design of these proving rings
is such that although it is necessary to take off the dial gauges to remove the rings for calibra-
tion, the stressed ring mountings remain undisturbed so that the calibrations are reliable.lz
The two tangential load proving rings were of slightly differing stiffness which must, in prin-
ciple, induce a small unbalanced force in the system. Shims inserted at a wheel/torque arm
contact during shear resulted in there being no change in the measured torque based on the
mean of the two proving ring readings. The distance from the main shaft axis to the wheel/
torque arm contact was determined by careful measurement at the end of the shear stage of the
test.
The accuracy of the normal loading system can be affected by inaccuracies in the lever arm
and by friction in the bearings. With the apparatus assembled, minus a soil sample, the lever
arm was balanced by adjusting the counterweight. Only 0.02 lb additional weight was re-
quired on the hanger to overcome static friction, which is negligible. The calibrated side-
friction proving ring was connected directly via a load transfer ring to the normal loading
system. For hanger loads less than 3 lb (0.2 lb/sq. in. on the sample) the normal stress was in
error by up to 3% but above this load the indicated error was only 0.3% and was constant
(Nowacki, 1967; Garga, 1970). This calibration was unaffected by lateral loads up to 60 lb
applied to the main shaft immediately above the torque arm. Hence, provided the vertical

12Recent in situ calibrations at I.C. by P. W. Davies show no measurable difference over a wide load
range.
A NEW RING SHEAR APPARATUS 291
loading system is balanced, no significant errors appear to arise in determining the normal load
and shear load transmitted across the soil on the plane of relative rotary motion.
The direct results of a ring shear test are corresponding values of the net normal load W, the
total torque M, and the angular displacement 8. In determining the stress-strain relation-
ship, and in particular the residual strength, a number of alternative assumptions are possible:
(a) that the normal stress u’,, and the shear stress T are uniformly distributed across the
plane of relative rotary motion;
(b) that the normal stress u’,, has a specified form of variation and the shear stress 7 has a
correspondingly similar distribution;
(c) that the normal stress u’” is uniformly distributed whereas the shear stress 7 varies
linearly with the radius (Cooling and Smith, 1936) ;
(d) that the normal stress u’,, is uniformly distributed whereas the shear stress 7 is a
function of the shear strain (Hvorslev, 1939; Payne and Fountaine, 1952; Nowacki,
1967).
Alternative a appears to be the most realistic choice. It is the simplest both in principle
and in application and assumes the soil to be plastic, as does b. Various possible stress distri-
butions under alternative b were analysed to determine their effect on the measured value of 4’
(Table 1 and Fig. 11). The method of analysis is given in the Appendix. Fig. 11 shows that
the error in assuming a uniform stress distribution is less than about f 2” for 4’ < 20” even for
the most extreme and unlikely non-uniform stress distributions considered. Alternative c is
theoretically untenable and is not discussed further. Alternative d, similar to c but requiring
a semi-graphical analysis of the moment-twist curve, is not amenable to direct calculation of
the possible errors. Analyses of tests by Hvorslev and Kaufman (1952),13 Nowacki (1967),13
and by the Norwegian Geotechnical Institute (196S),14 indicate that the effect on the value of
+‘I- is negligible although the shape of the stress-strain curve near C’p and the value of $lp may
be different from that obtained with a. Alternatives c and d both assume the soil to be
elastic which is less acceptable than a or b.
The average displacement is calculated on a mean sample radius of 25 in. The average
normal stress dn and average shear stress 7 calculated on assumption a are given by :
, W
un =
7T(r,2 - Y12)
3M
T= 2?7(r,3 - 113)
so that
3M(r, + ~2)
tan +’ = 2W(r,2+r,r2+r,2)

= 0.395; for y1 = 2-O in., r2 = 3-O in.

TYPICAL TEST RESULTS AND THEIR SIGNIFICANCE


Blue London Clay
An extensively fissured 130 ft thick bed of blue London Clay at Wraysbury, near Staines,
is overlain by about 15 ft of gravel and alluvium. The strength properties of the blue London
Clay from the nearby Ashford Common site were described by Ward et al. (1965) and
Bishop et al. (1965). More recently, Skempton and Petley (1967), Fookes and Parrish (1968),
and Skempton et al. (1969) have examined joints and fissures at Wraysbury and their influence

I3 The tests for which details are given were done in a ‘fixed ring’ type of ring shear apparatus (Hvorslev,
1939) which introduces other difficulties in interpretation.
I4 These tests were on soils with only a small decrease in 4’ from peak to residual.
Table 2(a). Results of ring shear tests at I.C. on blue London Clay from Wraysbury
LL=72%, PL=29%, CF=57%
X-ray diffraction at R.S.M. : quartz 30%, chlorite, lo%, illite 35%, kaolinite 15%, montmorillonite 10%
- /
Test Peak stresses T Residual stresses Strength parameters Average Average Average Shear- Approx. End of test
no. - initial final final ing
Normal Shear Normal Shear Peak Residual W.C., W.C., W.C. in rate! Elapsed z
ub, O’n, % % shear in./mm time,
b/sq. in. 1Lb/s:. in. lb/sq. in. b/s:. in zone, % :
days 0
- V
Re- .
mouldet 28.8 10.0 26.9 4.51 43.8 37.3 - 0~00015 6 6.6 CJ
33 .
RS-1
_- - m
.
Undis- c,
turbed m
RS-2 29.6 13.6 29.0 4.78 28.7 33.7 33.0 0.0003
RS-3 11.3 6.2 11.9 1.97 29.0 33.4 36.2 0~00015
RS-4 0.96 29.2 35.3 40.0 0~00015
RS-5 4z.i 1Z.i 4i.E 6.62 28.9 31.2 32.0 0.0003
-

Table 2(b). Results of ring shear tests at Harvard on blue London Clay from Wraysbury (La Gatta, 1970)
Sample size: o.d.=7.1 cm, i.d.=5.1 cm, thickness=0.2-0.3 cm, LL=72%, PL=22%
- - - -
Test/stage Peak stresses Residual stresses
5
_
Strength parameters Initial Shearing Approx. I End of test
‘_
no.
I I I
.- I W.C., rate, displace- s-
Normal Shear Normal Shear Peak Residual % cm/min ment Displace- Elapsed k
0’11, 7. u’n, ment, time,
kg/cm2 kg/cm2 kg/cm2 kg jc’ma (c’?o), d’=:rn cm days 2
--

Remoulded :
0.362 1.0 4.0 m
0.177 19.90 10.0” 67.0
2fl ;:; 0.660 2.0 0.293 18.30 63.4 1; :: z
212 4.0 4.0 0.572* 16 70 ;:;
3 8.0 2-180 8.0 I.175 l& 62.1 0.0048 4 32 4.0 $
Undisturbed 18.0”
0.315 1.0 0.160 17.5” I 27.7 40 60 7.4 z
0.685 2.0 0.339 18.9’ 28.0 35 55
1.322 4.0 0.650 18.3” 28.5 3 ;:; u
2.560 8.0 1.320 17.7” J 25.3 10 :z 6.8 *
- - Ez
* Rate of displacement reduced tenfold at end of this test: +=0.561, +‘,=&+O”. 0
A NEW RING SHEAR APPARATUS 293
on shear strength. Agarwal(l967) has determined the drained and undrained shear strength
of the clay from Wraysbury.
One ring shear test on remoulded clay and four on undisturbed specimens (Table 2(a)) were
performed on samples from Wraysbury from pits at a depth of about 5 ft below the top
of the clay stratum. The stress-deformation relationship for test RS-2 (Fig. 12) is shown in
terms of the stress ratio ~10’~and for the residual state, +‘r= tan-l (~10’~)where c’=O. This
sample was subjected to an average displacement of 52.3 in. under a nominal (T’~= 30 lb/sq. in.
The initial shear rate l5 of 0.0003 in./min was chosen to achieve full dissipation of excess pore-
water pressure. During test RS-2 the rate was varied by a factor of x 0.5 in order to study rate
effects and by factors of x 125 and x 625 to obtain large displacements rapidly. All the present
ring shear tests showed that for the slow rates used the residual strength is relatively insensi-
tive to the rate of shearing, in agreement with Petley (1966)) Tchalenko (1967), Kenney (1967)
and Skempton and Hutchinson (1969). After peak T/U’,,the confining ring gap may be closed
(to minimize sample loss by squeezing) and the rate of shearing increased, in order to decrease
the testing time. When the desired large displacement has taken place the gap then can be
reopened and the shearing rate reduced to its initial value. Usually one day was allowed for a
new state of equilibrium to be reached and for any excess pore pressures to dissipate before
readings were taken.
The readings were normally taken with a 0.003 in. nominal gap between the upper
and lower confining rings. However, it was observed that the gap, once opened, does
not necessarily remain uniform as shearing proceeds. Development of the slip surface
often results in uneven vertical deformation of the sample and the upper loading platen and
confining rings will tilt slightly until restrained by the loading mechanism. This would intro-
duce a friction component if metal to metal friction occurred at any point or if squeezed-out
clay was jammed in the gap. Enlarging the apparent gap periodically by 0~001-0~002 in.
provides a check as to whether friction losses exist or not, e.g. at displacements of 1.0, 2*7,3.5
and 16.2 in. in test RS-2. Correct r/o’,, values can only be obtained with the confining ring
gap open.le The vertical deformations of the sample are consistent but subject to some
inaccuracy since they are affected by loss of clay by squeezing through the confining ring gaps.
Test RS-2 was stopped for two days at a displacement of 3.9 in. to see if the blue London
Clay was capable of ‘healing’. No gain in strength was observed when shearing was re-started.
Hvorslev (1939) found a 20% increase in the residual strength of Vienna Clay in a ring shear
test stopped for l& h, but this may have been due to excess pore-water pressure set up by
shearing too rapidly.
In test RS-1, and in all later tests, it was observed that differing loads were registered on
the two tangential load proving rings. Only a small portion of the difference could be attri-
buted to the fact that the two rings were of unequal stiffness. By inserting shims of appro-
priate thickness between the torque arm and the wheel of one proving ring, equal loads were
obtained in each ring, while the average load (and hence T/~‘,J remained unaffected. The
magnitude of the initial disparity between the two rings before shims were inserted was not
consistent in all tests and was inferred to be due to slight differences in setting up the apparatus
and in the mode of failure of the individual samples.
The polished, striated slip surface of the remoulded sample after testing is shown in Fig. 13.
The upper portion was removed with a spatula by sliding off short sections, not without diffi-
culty. A single, well-defined slip surface was observed in the plane of the confining ring gap.
The undisturbed sample, test RS-2, exhibited two noteworthy features. Two polished,

15 The shear rate for all the ring shear tests referred to in this Paper is defined with respect to the mean
sample radius, i.e. 2x5 in. in this case.
I8 Similar problems arise in a conventional shear box test where control of the gap between the two
halves is much more primitive. Approximate corrections for friction effects are given by Gibson (1951) and
Petley (1966).
294 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J

b k
A NEW RING SHEAR APPARATUS 295
striated, parallelslipsurfaces (Fig. 14(a)) were discovered about 0.05 in. apart, indicating major
displacements on more than one surface. A secondary highly polished inclined surface (or
fissure) emanating from the tip of a radial fin (Fig. 14(b)), was found. There was no visible
evidence that the blade had been displaced laterally in the clay so that movement on this in-
clined surface must have been very small indeed. Similar features to those shown in Figs 13
and 14, but of varying extent, were observed in the remaining three tests. However, the
principal displacement is clearly in the plane defined by the junction between the two pairs of
rings. Some details of the shear zones examined under both the optical and electron micro-
scopes are described by Garga (1970).
Figure 12 clearly shows a marked drop in T/c’~ from the peak at less than 0.1 in. displace-
ment to $‘= 12” at 1.0 in. displacement. However, continued shearing results in a reduction
to Cfr=9*4” beyond about 15 in. and is subsequently constant up to 52 in. displacement.
Similar behaviour was observed in the other four tests although the displacement at which C’,
was established varied from 5 to 20 in. It is evident from Fig. 12 that it is not easy to decide
whether, if the test had been continued to say 100 in. displacement, the residual strength +‘r
would have been reduced significantly below 9.4”. Test E-2 is therefore replotted on a semi-
logarithmic displacement scale in Fig. 15 which shows that beyond about 15 in. displacement
the curve is horizontal, indicating that the residual strength is estab1ished.l’ Also shown in
Fig. 15 are two drained multiple reversal direct shear box tests on blue London Clay from
Wraysbury (Agarwal, 1967). The undisturbed direct shear box sample exhibits a very
erratic curve which does not satisfy the definition of residual strength. These curves, typical
of a multiple reversal direct shear box test, show an increase in apparent residual strength with
increasing displacement, which is unacceptable. The ‘troughs ’ in the stress-displacement
curves on the third and fourth travels agree fairly closely with the ring shear data. This
appears to be fortuitous in the present tests and does not occur in other similar tests, e.g.
Fig. 16. In any case, there seems little theoretical justification for taking the multiple reversal
direct shear box test ‘troughs’ as the residual strength. Tests on a slip surface found at the
site after a failure during construction (Fig. 15) show near-horizontal curves on both the first
and second forward travel which indicate a #‘= value very close indeed to the value of 9.4” given
by the ring shear test. This test is one of two similar tests at u’,, = 15 and 30 Ib/sq. in. for
which $‘1=8*2” and lOal”, respectively, the residual shear stress being taken as that at the
end of the first travel. In both tests the value of +lr at the end of the second forward travel,
8.5” and 10*2”, respectively, was greater than that during and at the end of the first forward
travel. It appears that the reversal process had disturbed the particle orientation along the
slip surface and the residual strength should be considered to operate at the end of the first
travel. Both tests give 41rvalues consistent with the ring shear tests and not with the multiple
reversal direct shear box tests on undisturbed samples.
Multiple reversal direct shear box or cut-plane triaxial tests (Table 3) all give much higher
values of residual strength for blue London Clay than is obtained in the ring shear test.
Petley’s results for two sites, $‘p= 12.5”-14*7”, are consistently high compared with the present
value of +‘==9.4”. This difference cannot be explained by differences in index properties or
clay fraction. The two cut-plane triaxial tests on Ashford Common Shaft material for which
+‘r values of 1I-6” and I1 -5” were obtained suggest that even artificial polishing of slip surfaces
by a spatula or glass plate does not establish maximum particle orientation.
A manual ring shear apparatus was made in which an undisturbed sample was presheared
through very large displacements very rapidly in order to produce an artificial slip surface.
A 12 in. dia. x 9 in. high sample of blue London Clay from Wraysbury, previously consolidated
to 60 lb/sq. in. in a large dissipation cell was confined in a divided steel ring and loaded axially
through roughened steel platens. The upper ring was manually rotated 12 times relative to

I7 This method of presenting the results has also been suggested by La Gatta (1970).
Table 3. Results of residual strength tests on blue London Clay: direct shear box and triaxial tests
-. - - -_
Location Sample and test type Normal Shear stress, T, Residual strength hverage Reference
stress, / lb/sq. in.
I
I- -__ initial rate! -
~6, Average W.C., in./mm
lb/sq. in. Peak Residual (& 0) +‘P %
Ashford ( Iut-plane drained triaxial test
Common Undisturbed : 641 , - - 11.6’ 11.5” 25 Bishop
Shaft LL=71, PL=29, CF=53 1301 es - - 11.5O et al. (1965)
-_ - __ __
Herne Bay I Drained, multiple reversal direct
shear box test
Undisturbed: 17.7 14.6 4.0 12.7”
LL=81, PL=33, cFz61 31.2 15.6 66-7.2 11~9”-13~0” 13.5” 27
44.7 31.2 11.0-12.5 13@‘-15.6” )~00004
- __ __
Slurry : 13.2 12.0° Petley
LL=85, PL=34, CF=59 26.7 1i.i 5.72:6 12.1”-13.9” 12.5’ 62 (1966)
35.7 11.2 6.8-7.8 low-12.30

(M-plane drained triaxial test


Undisturbed : 25.0 - 6.7 15.00 14.7O 27 I.000024
LL=81, PL=33, CF=61 43.4 - 11.1 14.3”
___ __
Leigh on 1Drained, multiple reversal direct
Sea shear box test
Undisturbed : 17.7 5.0 15.80
LL=67, PL=26, CFz51 31.2 7.5-8.5 13.5”-15.2” 14.5” 27
44.7
j55”11.0-12.0 13+3°-15~00

Slurry : 10.2 4.5 18.4” Petley


LL=70, PL=26, CF=49 22.2 9.5 ;:; 14.2” 14.5” 55 (1966)
31.2 11.5 a.2 14.70
___ __
(lut-planedrained triaxial test
Undisturbed : 25.0 - 15.0” 14.70 27
LL=67, PL=26, CF=51 49.6 - 14.4”
_-
Wraysbur) II Drained triaxial test
Undisturbed : presheared to
large displacement : 66 17.2 12.2 10.50 10.50 29 Garga
LL=72, PL=29, CF=57 (O’s = 55) (1970)
Wraysbury Drained, multiple reversal direct
shear box test
Undisturbed: 7.5 17.1°
LL=70, PL=29, CF=58 15.0 3:; 14.2”
22.5 5.3 13.3”
30.0 9.0 16.7” 13.5” 28 0~00012
45.0 Agarwal
10.0 12.5’
60.0 15.3 (1967)
14.3”
75.0 17.0 12.8”
90.0 17.3 10.9”
I- -- _-
Field slip surface: 2.8 8.2’ 9.2” 31 0~000054
LL=74, PL=30, CF=60 6.0 ;:; 10.1°
I !- I I -
Direct shear box test : sample size, 6 x 6 x 2 cm thick. Triaxial test: sample size, 18 in. dia. x 3 in. high.

Table 4. Results of ring shear tests at I.C. on brown London Clay from Walthamstow
LL=66%, PL=24%. X-ray diffraction at R.S.M.: quartz 22%. feldspar 3%, plagioclase 4%, calcite 2%. dolomite 5%, illite 3l%,kaolinite 15%,
montmorillon~te 17%
- - -
Test/stage Peak stresses I Residual stresses Residual Shearing I End of test Average Clay
no. - _- strength rate, - initial fraction,
Normal Shear Normal Shear in./min Displace- Elapsed W.C.,
%
06, 7. a’., (“$= r>O) ment, time, %
lb/sq. in. lb/sq. in. lb/sq. in. lb/s:: in. in. days
-- -- __ __ __
Undisturbed
l/l 31.8 14.4 31.0 4.77 14.1
- - 16.0 2.51 ;:;I 155
l/2
l/3 - - 8.6 1.46 . 0 17.7
l/4 - - 0.91 1E 0.0003 19.8 31 53
l/5 - - 2g.t 3.51 8.4” 24.8
l/6 - - 36.3 5.10 8.0” 29.0
‘I7 - - 4.5 0.84 10.6” 31.3
l/8 - - 2.3 0.50 12.20 33.5
l/9 - - 1.0 0.25 13.8” 38.0 110
-_ _- _-
Slurried
2/l 13.1 6.06 13.1 2.25 7.0
212 - - 168 1;:;: 0~00015 8.2 41 64
213 - - 1i.i: 2.30 9.6” 9.7 40
- -. - - - - - -
298 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN
0.5

l Undisturbed: drained ring shear test RS- 2,


nominal 0; =30 Ib/rq.in., Garga (1970)

0 Undisturbed; multiple ~reversal, drained direct shear-box tesr


CA=30 Ib/sq.in.,Agarwal (1967)

A Drained direct shear-box test with one rwersal an slip surface


s;= 30 Ibhq.in., Agarwal!1967)

1% 2nd ccc. denotes no. offorward travel

0
0.1 0.5 I.0 5 IO
Displacement in inches
Fig. 15. Development of residual strength with increasing displacement: blue London Clay from
Wraysbury, test RS-2

0.5
Blue London Clay from Wrayrbury

l Undisturbed: drained ring shear wrt US-5,


\
nominal ‘JA -41 Ib/rq.in., Garga (1970)
\
\,
04 a Undisturbed; multiple reversal, drained direct shear-box test
CA= 45lb/sq.in., A&al (1967)

Ist, 2nd etc. denotes no. of forward travel

0.3 -_ -

z
r a,
1st ‘b

0.2 -

0.1

Fig. 16.
O-
0.1

Development
I.0
Displacement in
of residual strength with increasing displacement:
Wraysbury, test RS-5
5
inches
IO
-

- 1
50
blue London Clay from
A NEW RING SHEAR APPARATUS 299
the lower ring while maintaining an average normal stress of 55 lb/sq. in. in a 100 ton triaxial
loading frame. Three 14 in. dia. sampling tubes inclined at 55” to the horizontal were pushed
into the block to obtain samples including the shear plane(s) for drained triaxial tests.
Unfortunately, of the three samples taken, one was destroyed on extruding, while the second
failed along two wedges which made the analysis unreliable. The third (Table 3) gave a value
of 4’ == 14.6” and C’*= 10*5”, assuming c’=O in both cases.
The strength-effective stress relationships for the ring shear tests together with other
shear strength data on blue London Clay from Wraysbury are shown in Fig. 17. Agarwal’s
peak strengths from vertically-orientated triaxial and direct shear box tests on undisturbed
horizontal samples are remarkably consistent, c’=4.5 Ib/sq. in. and 4’ =20*5”.r* The peak
strength from the ring shear tests on undisturbed samples, c’=2.0 lb/sq. in. and $‘=19.7”,
shows a reduction in c’ but essentially no change in (6’. The residual strength of undisturbed
clay obtained in the ring shear tests, c’=O, +‘,=9*3”, is much lower than that obtained in
multiple reversal direct shear box tests, c’=O, $‘r = 13.5”, and the data are much less scattered.
The value of 4’ is in all cases, and in particular for the ring shear residual strength, independent
of the stress level. This, coupled with the lack of scatter, suggests that the experimental
errors are small.
The single ring shear test on blue London Clay remoulded at a water content of 43*80/
gave a value of #‘r=9*50, agreeing almost exactly with the average, +‘r=9*3”, for the four
undisturbed ring shear samples. Taking c’ =0 for the remoulded clay a peak value of 4’ =
19-l’ is obtained. It will be noted that this is close to the peak value of 4’ obtained for undis-
turbed clay using this size of sample. The value of 4’, of 10.5” obtained from the drained tri-
axial test on an undisturbed presheared sample lg is in reasonable agreement with the ring
shear test results, but suggests that this method can lead to a small over-estimate of the resi-
dual strength.
Substantial vindication of the new ring shear apparatus is provided by independent tests
on blue London Clay from Wraysbury by La Gatta (1970) at Harvard with a new annular
shear device of a different design. In the Harvard apparatus a ‘smear’ sample, 2-3 mm thick,
o.d. = 71 mm, i.d. =50 mm, confined by solid PTFE rings, was tested. After sufficient dis-
placement had been applied, the outer confining ring was temporarily lifted so that friction
there was eliminated, while minimizing the effect of loss of clay by squeezing on the normal
stress distribution.20 Th e results of tests on remoulded and ‘undisturbed’ samples (Table
2(b) and Fig. 18), are almost identical to those obtained by Garga (1970). La Gatta’s re-
moulded samples were mixed with distilled water to the liquid limit and air-dried to a lower
water content, whereas Garga’s single sample was remoulded with the addition of distilled
water but without subsequent air drying. The average residual strength obtained by La
Gatta in four tests on ‘undisturbed’ samples was 9*3”, the envelope being linear through the
origin and with very limited scatter, as compared to Garga’s +‘r=9.3” and similar scatter.
The addition of distilled water and air drying appears to have consistently reduced the value
of +‘r by 1-O”in each of the four tests. The nature of the apparatus and the sample tested by
La Gatta is such that, in contrast to the I.C./N.G.I. apparatus, it is not possible to measure an

I8 The peak effective stress parameters of vertically and horizontally-oriented triaxial samples, as deter-
mined by Agarwal(1967), were very similar whereas samples inclined at 45” were apparently slightly weaker,
c’=4.5lb/sq. in., 4’= 19.0”. The agreement between the effective stress strength parameters obtained with
the vertical triaxial samples and with the horizontal shear box samples is hence subject to qualification.
I9 The analysis of triaxial tests in which failure occurs on asingle inclined slip surface is difficult onaccount
of uncertainties concerning the restraining effect of the sample sheath (La Rochelle, 1960, 1967; Webb, 1966;
Chandler, 1966; Blight, 1967).
ao No provision was made to lift the inner confining ring but since the effect of lifting the outer ring was to
reduce the residual strength of the blue London Clay by only 0.4”, the lack of this facility seems likely to be
unimportant.
300 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN
Undisturbed: drained rinq sheor
tests at Imperial Collcqe,vr-29

Undisturbed: drained rinq theor


tests .a, Harvard, uc.rZB
Remoulded:droined ring rkor
tests at Harvard, nx.=bl

00 120
Normal effective stress. lb/+.

Fig. 18. Strength-effective stress relationships for blue London Clay from Wraysbury: ring shear tests at Imperial College andHarvard
302 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

undisturbed peak strength because the ‘smear’ sample must fail at the sample/platen interface
where it is already remoulded as a result of preparation. The ‘undisturbed’ peak strengths
measured by La Gatta are, more correctly, remoulded peak strengths and are in good agree-
ment with the single remoulded sample tested by Garga. They give 4’ = 18.0” compared to a
peak value of +‘= 19.7” measured at I.C. on the undisturbed samples. There appears to be
little difference in La Gatta’s tests between the peak strengths of the two types of sample
although the three remoulded sample points are less definitive. In test 2 (Table 2(b)), a
second loading stage was imposed. The measured residual strength was almost identical to
that obtained in the first stage and when the rate of displacement was reduced tenfold the
reduction in +‘r was only 0.2”.

Brown London Clay


Samples of weathered brown London Clay were obtained from a site in Walthamstow
adjacent to Banbury Reservoir on the River Lea. The strength properties of the brown
London Clay from other sites have been studied by Petley (1966) and Skempton and Petley
(1967), and from Walthamstow by Petley (1969). T wo tests, one each on undisturbed and
slurried clay, were conducted. The undisturbed sample was taken from a depth of 3.5 ft below
the top of the clay. Numerous slickensides and shear planes were found at the site, evidence of
past instability probably caused by valley erosion and undercutting by the River Lea and by
solifluction (Hutchinson, 1969).
The stress-deformation relationship for test 1, on undisturbed clay, is shown in Fig. 19.
The test was conducted by shearing the same sample under a series of different normal stresses.
After the residual strength under one normal stress had been determined, the motor was
stopped and the new normal stress was applied. The sample was then allowed to consolidate
(or swell) under the new load, for at least two days before shearing was re-started. It is
apparent from Fig. 19 and Table 4 that the sample showed a marked variation in residual
strength as the effective normal stress was altered. When the sample is sheared under a re-
duced normal stress a peak is first observed and the difference between the peak and the
residual strengths increases as the effective normal stress decreases. When the normal stress
is increased no such peak occurs. If this phenomenon is repeated in nature it suggests that the
shearing resistance to be overcome when re-initiating movement on a pre-existing slip surface
may exceed the residual value if the movement is associated with a decrease in effective normal
stress. This will influence the interpretation of field behaviour in highly plastic clays such as
the London Clay.
The vertical deformations in test 1 (Fig. 19) reflect the value of the normal stress u’~ and,
as with the blue London Clay, are affected by loss of clay due to squeezing.
The slurried sample was prepared from air-dried clay, ground to pass a 100 B.S. sieve,
mixed with water up to a water content of 160% and consolidated under 250 lb/sq. in. in a
9 in. oedometer. The clay was unloaded and allowed to swell before cutting a ring shear
sample from it and conducting a three-stage loading shear test (test 2, Table 4).
The residual strengths for each loading stage of test 1 show an almost unique relationship
when plotted against effective normal stress (Fig. 20), independent of the loading sequence.
There is evidence from a semi-log stress-strain plot that at the end of the first stage the residual
strength had not quite been established, which accounts for this point lying above the curve
through the remaining nine points.
The three points for the slurried sample lie on a similar curve but slightly above the undis-
turbed sample, possibly owing to alteration of the soil or pore water chemistry as a result of
air drying. The opposite effect was observed by La Gatta on blue London Clay (see Fig. 18).
The important conclusion to be drawn is that residual strength is independent of stress history
since the points fall on a unique curve dependent only on the magnitude of the effective normal
Table 5. Results of residual strength tests at I.C. on brown London Clay: direct shear box and triaxial tests
-
Location Sample and test type Normal Shear stress, r, Residual strength iverage Shearing Refer-
stress, lb/sq. in. initial rate, ence
- Average
ah,
lb/sq. in.
r Peak Residual
r- fl=
W.C.,
%
in./min.
__ __ .- -- --
Hendon Irained, multiple reversal direct 4.0 4.3 1.5 20.5”
shear box test. Undisturbed: 5.0 4.8 2.0 218”
LL=82, PL=33, CF=60 8.0 6.5 18.0”
12.0 7.4 42:; 19.30 l/=0.5
13.2 9.2 22.4” +‘r = 14.6”
~‘~2.6
17.7 4.cz4 12*7”-14.0” 29
26.7 .$‘=23.0”
1Z.i 6.6-7.0 13.9’14.7”
26.7 15.0 16.7’
31.2 16.8 8.csO 14.4”-16.1”
40.2 15.9 8.5 11.9O
40.2 21.5 9.0 12.6’

Slurried :
LL=Sl, PL=34, CF=61
_
44.7

13.2
13.2
--
21.4 109-12~0
--
12.6”-15.0”

15.3”
15.3”
-_
I Petley
(1966)

22.2 14.6” l/=0.8 59


22,2 125” 8b’,= 11.3”
31.2 1 13.0”
31.2 1 12.10
-- --
:ut-plane, drained direct shear 1.5 14.0”
box test. Undisturbed : 1Kl - 2.3 12.9” c’=O.3 29
LL=82, PL=35, CF=58 15.0 - 3.0 11.3” b’, = 10.3”
20.0 - 4.1 11.6”
-- __ -_
Zut-plane, drained triaxial test. 19.0 - 15.6” b’, = 14.8” 29
Undisturbed 36.9 - 14.00
_- -- -- --
Herne Bay Irained, multiple reversal direct 26.7 10.2 12.20 L’, = 13.0” 28
shear box test. Undisturbed: 44.7 21.7 13.8”
LL=79, PL=30, CF=66 _-
Slurried: 17.7 6.0 11.8” 60
LL=79, PL=30, CF=66 35.7 10.8 0~00004
10.9”
:ut-plane, drained direct shear
__ Petley
box test. Undisturbed: 22.2 4.4 28 (1966)
LL=82, PL=30, CF=68 35.7 7.2
_- --
X-plane, drained triaxial test, 35.9 12.4’ 0~000012
Undisturbed 44.1 ;:; 12.20 28 0.000025
53.5 11.2 11w 0~000012
- -2 -
Brentwood 1Drained, multiple reversal direct 1 17.7 I 10.4 I I 15.2” I
shear box test. Undisturbed : 31.2 1 14.6 6.8Y.612.3”-13.7O fr = 13.9” 30
LL=72, PL=29, CF=52 44.7 17.9 10.0-l 15 12.6”-14.4”

Slurried : 13.2 5,2 3.6 15.2’ D.00004


LL=77, PL=29, CF=53 26.7 11.0 11w fr= 12.9” 53
35.7 12.1 7.oYo ll.l”-12.6” 1Petley
Zut-plane, drained direct shear (1966)
box test. Undisturbed : 17.7 - 3.5 11.2” &,=ll.l~ 31
LL=73, PL=29, CF=52 31.2 - 6.7 12.10
--
:ut-plane, drained triaxial test. 23.9 - 12.0” fqp= 13.00 30 D~00002
Undisturbed 36.9 - 14.0” 0.000025

Guildford Irained, multiple reversal direct 4.2 - 17.2”


shear box test. Slip surface: - 15.1”
LL=83. PL=32, CF=57 ;:; - 17w
10.4 - 14.0° c’=O.4
15.0 - 12.1” +‘,=11.6” :jkemp-
15.0 - 16.0” 34 D~00004 ton
20.0 - 12.4’ or and
20.0 - 10.8’ Petley
22.0 - 13.3” $‘r = 13.8” (1967)
25.0 - 13.7”
31.2 - 11.6”
35.0 - 11.9O
--
Waltham- Irained, direct shear box LL = 76 8.7 15.4”
stow test. Slip surface: PL=30 13.2 15.6’ 32
CF=68 17.7 12.7’
26.7 12.9O

LL=43 14.2”
PL=21 1i.Z ;:; ;:“3 14.0” lyr= 14.0” 38
CF=40 26.7 6.4 12.9” 0~00012
35.7 8.8 g:: 13.4”
-~-
LL=72 2.5 2.1 13.6”
PL=27 14w 34
CF=65 ;:; g:; Petley
26.7 14.50
(1969)
35.7 9.1 9.0 14.2”
-
Z&-plane, drained direct shear 13.6”
box test. Undisturbed : 1% ;:; 14G3” 32
LL=70, PL=26, CF=50 26.7 7.3 13.7O yp = 14.2’
35.7 9.6 14.6” 0~000024
i
Drained triaxial test. Slip surface: 18.3 - 14.10 fr= 13.70 31
LL=71, PL=26, CF=63 37.4 - 13.2’

Direct shear box test: sample size, 6 x 6 x 2 cm thick. Triaxial test: sample size, 14 in. dia. x 3 in. high.
306 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

v .rv
I I I I
lb’
Ring shear tests on brown London Cloy from Walthamstow

o Undisturbed sample, w.c.=31, LL=bb, PL=24, CF=53 15’

Slurried sample, w.c.=4I, LL=68, PL=25, CF=b4

Ist, 2nd etc, denotes loadinq stoqe series no.

Effective normal stress, fit lb/sqin

Fig. 20. Variation in residual strength of brown London Clay with


stress level

stress. The residual strength of the brown London Clay tested varies from 14” at u’,= 1 lb/sq.
in. down to 8” at ~I,=36 lb/sq. in.;21 in particular the increase in residual strength below about
10 lb/sq. in. is very marked. The practical significance of this as applied to slope stability
analyses is discussed later.
It is not certain why the brown London Clay shows a variation in 41r with ufn, whereas the
blue London Clay does not. Confirmation of the reliability of the multistage loading tech-
nique is provided by the results for Studenterlunden Clay discussed later. According to
Kenney (1967) soils with a high montmorillonite content show a variation of $‘, with u’~
whereas soils composed of heavy non-clay minerals do not. X-ray diffraction analyses of the
blue and brown London Clay (Tables 2 and 4) show that the latter has a higher montmorillonite
content, but whether this difference is significant is speculative. The curve for the brown
London Clay is not asymptotic to that of the blue London Clay but shows a lower residual
strength at high stresses. To what extent ring shear tests on brown and blue London Clay
from different sites will show +lr-- u’~ relationships identical to the present limited tests is un-
certain. The self-consistency of the present tests is very promising and suggests that any
differences should be readily determined from ring shear tests.
The residual strength of brown London Clay obtained from
(a) multiple reversal direct shear box tests on undisturbed or slurried samples, or
(b) cut-plane direct shear box tests on undisturbed samples, or
(c) direct shear box tests on the slip surface, or
(d) cut-plane triaxial tests on undisturbed samples,22 or
(e) triaxial tests on the slip surface22
on clay from five sites including Walthamstow all give residual strengths 2-6” higher than those

21 It is interesting to note that Langer (1938) (see also Hvorslev, 1939, Fig. 12) obtained a value of
+‘= = 8.8” at a normal effective stress of 90 lb/sq. in. in a torsion shear test on remoulded London Clay.
2a In these last two tests the cut-plane or the slip surface was oriented at about 55’ to the horizontal, and
the top loading platen was unrestrained laterally, following Penman (1953).
A NEW RING SHEAR APPARATUS 307
measured in the present ring shear tests (Table 5 and Figs 21 and 22). The multistage ring
shear test on undisturbed clay shows (Fig. 21) a smooth curved envelope, clearly passing
through the origin with the slurried sample in very close agreement. Also shown are tests by
Petley (1966) on brown London Clay from Hendon. The peak strength of this clay obtained
from the first forward traverse of a direct shear box test is c/=2.6 lb/sq. in., +‘=23.0” com-
pared with c’=4*5 lb/sq. in., #‘=205” obtained for Agarwal’s tests on blue London Clay
(Fig. 17). The difference in 4’ is marginal in view of the scatter of the data whereas the co-
hesion intercept is lower for the weathered (brown) clay, as is to be expected. The peak
strength obtained during the first loading of ring shear test 1 lies close to the peak for the Hen-
don samples. The residual strengths from the four types of test on the Hendon samples are
generally at variance with each other. The multiple reversal direct shear box tests on undis-
turbed clay give c’=O*5 lb/sq. in., #‘r= 14.6” between 4 and 30 lb/sq. in. effective normal stress,
with some indication of curvature at higher stresses. The two cut-plane triaxial tests also lie
on this line. However, the slurried samples, c’=O3 Ib/sq. in., $‘r = 1 l-3”, show a reduction in
residual strength and the cut-plane direct shear box tests on undisturbed clay give even lower
values, c’=O*3 Ib/sq. in., $‘r= 10.3”. AI1 of these tests on Hendon clay show variable cohesion
intercepts which one would expect to be smaller, if not zero, at a true residual state.
The ring shear tests are also compared with similar tests by Petley (1969) on brown London
Clay from Walthamstow, i.e. the same site. In contrast to the Hendon data the three types of
test performed by Petley all give the same residual strength, +‘r=14*O”, and no cohesion
intercept. Again this is greatly in excess of the residual strength obtained from the ring shear
tests. Additional data, all supporting this view, are given in Table 5 for brown London Clay
from Herne Bay, Brentwood and Guildford.
The lack of self-consistency between the various tests leaves the engineer in a dilemma in
deciding which, if any, is the correct result. It seems likely that the limited ring shear data are
more reliable since the results are more self-consistent and less subject to objections in analysis
and interpretation present in the other types of test.

Weald Clay
An undisturbed sample of weathered grey Weald Clay from a bedding shear zone at Arling-
ton near Eastbourne was subjected to a multistage ring shear test. Typically, Weald Clay is a
stiff, fissured, heavily over-consolidated material. The sample which resembled damp soap
flakes, consisted of a mass of tiny clay lenses typically 5 mm x 5 mm x 1 mm thick, lying parallel
to each other and with a dull surface. The residual strength of weathered Weald Clay from
Sevenoaks is discussed by Skempton and Petley (1967) and multiple reversal direct shear box
tests on material from Arlington were performed by Petley (1969).
The test was carried out in a similar manner to the tests on the brown London Clay and the
results (Table 6 and Fig. 23) indicate a very similar pattern of behaviour. During stages 1, 2
and 4, the confining ring gap was closed and the rate of shearing increased from its standard
value, 0.0003 in./min, to OX1075 in./min to obtain a large displacement more quickly, after
which the rate was reduced to its former value and the confining ring gap reopened before
taking readings. The 40 in. displacement required during the first loading stage to reach
residual is considerable. During subsequent stages the displacement required was much less
since the clay is effectively ‘presheared’. There is evidence from a semi-log plot to suggest
that the residual had not quite been achieved even at 40 in. displacement and that a further
reduction of 0.5” is possible. Similarly a reduction of 0.5” in the residual strength obtained
during the fourth and fifth loading stages is possible, but is less likely.
The residual strength of the Weald Clay varies from about 15” at u’,=6 lb/sq. in. down to
9.5” above u’,=60 Ib/sq. in., the increase in +‘= below u’~-20 - lb/sq. in. being rather marked.
The residual strength is independent of the stress history, i.e. loading sequence, since the
points lie on a unique curve.
Peak Residual
Multiple reversal drained
Q 0 Hendon
direct shear box tests
Undisturbed
Cut-,plane drained direct shear
w-30, CF-5( a
a box tests
LL=82, PL=3, 4
+ Cut-plane drained triaxial tests
Petley (1966)
Slurried: multiple reversal drained
+ 0
direct shear box tests on clay
from Hendon
wcr59, LL=81,, PL=34, CF = 61

Undisturbed: drained ring shear test I


0 Q on clay from Walthamstow

+ l
w.c.=31, LL=66, PL=24, CF =S3
Slurried: drained rinq shear test
‘on clay from Walthamstow
Carqa (1970)
I
w.c.=41, LL=68, PL=25, CF =64 J
I
I I I

cqn’, Normal effective stress, lb/q+.


Fig. 21. Strength-effective stress relationships for brown London Clay
Peak Residual

Drained direct shear box


0
on slip surface
- Cut-plane drained direct shear
0 Pctley (1969)
box tests on intact clay
Drained triaxial tests on
q
slip surface

Undisturbed: drained ring shear


0 0 test on clay from Walthamstow
-
w.c.=3l,LL=66, PL=24,
/d
Slurried: drained ring shear test /
+ .
on clay from Walthamstow 1;
w.c.=41, LL=68, PL=25, CF=64

IC, 15 20 25 30 50
0;;: Normal effective stress, lb/s5

Fig. 22.
1
Strength-effective stress relationships for brown London Clay from Walthamstow
310 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

0.30
I I I
- lb0
Rinq shear test on Weald Clay from Arlington near Eastbourne
Undisturbed sample from shear zone - 15”

w.c.=25, LL=bS, PL-32, CF=52


0.25

0.20

T
-, - IO”
05
7th
- 9’
0.15
- 8”

- 7’

- 6”
20 40 b0 80 I 1 120 140
Effective normal stress, 0%: II

Fig. 23. Variation of residual strength of Weald Clay with stress level

Table 6. Results of ring shear tests at I.C. on Weald Clay from Arlington
w.c.=25%, LL=65%, PL=32%, CF=52%
- -
Test/stage Peak stresses Residual stresses Residual Shearing / End of test
no. strength rate,
Normal Shear Shear in. /min Displace- Elapsed
a’,, ‘“$= r O) ment, time,
lb/sq. in. lb/s;: in. lb/s; in. in. days

Undisturbed
shear zone*
l/’ 8.9 3.38 8.9 2.20 13.90 40.7
l/2 - - 17.5 3.72 12.0” 48.0
l/3 - - 363 6.92 10.7” 50.0
l/4 - - 9.7 2.31 13.4” 0.0003 57.3
l/5 - - 1.72 15.40 58.6
l/6 - - IO.5 60.2
- - 19.3 ,“:,“: 62.5
l/7

* From Pit 2
Table 7. Results of residual strength tests at I.C. on Weald Clay: direct shear box and triaxial tests
-
Location Sample and test type Norma: 1 Shear stress, 7, Residual strength Average Shearing Refer-
stress
, lb/sq. in.
1.
initial rate, ence
f3n> T Average W.C., in./min
lb/sq. inI. Peak Residu al (Cf20) $‘= %
_I_
Sevenoaks Drained triaxial test. 1.9 16.80
Slip surface : 1Z 4.1 17.6” c’= 1.0 27
Pit 317H 25.4 14.4” +‘p = 12.8”
LL=69, PL=27, CF=71 52.2 1Z.X 13.1” ;kemp-
88.6 20.0 12.7’ ton
3.00005 and
Pit 300H 6.6 2.2 18.4” Petley
LL=62, PL=26, CP=50 14.4 18.4” c’= 0.7 24 (1967)
25.8 4:; 16.4” f$‘r= 14.9’
46.8 13.2 15.70
68.7 19.1 IS.50
-I-
Arlington Drained, multiple reversal direct
shear box test. 15.6 7.2 11.3”
Undisturbed : 31.2 11.3 10.7” 30
Pit 3. 46.8 16.9 11.00
LL=75, PL=29, CF=74 62.4 21.8 11.00
c’= 1.9 I-
-/
Pit 2. $‘= 18.2”
15.6 7.8 3.8 13.7”
LL=73, PL=30, CF=45 31.2 11.7 7.0 12.6” c’=O.9 27
46.8 18.5 10.6 12.8” 4’= = IO.50
?etley
62.4 23.0 12.3 11.1” I*00012
(1969)
Shear zone : 7.8 4.8 20.4”
Pit 2. 15.6 13.3O
LL=75, PL=30, CF=53 31.2 ;:; 11.90
46.8 13.3 c’= 1.9 11.40 29
62.4 16.5 ,$‘= 13.3’ 10.30
5.0 19.8”
7.8 ;:i 15.1”
10.0 4.4 15.6”
- -L

Direct shear box test: sample size, 6 x 6 x 2 cm thick. Triaxial test: sample size, 13 in. dia. x 3 in. high.
Fig. 24 (above). Strength-
effective stress relation-
_
ships for Weald Clay
from Arlington
Drained ring shear trrl on cloy from
0 shear zone at Arlinpton
we.- 24, U-62, Pl-Pb, CF=SO

Fig. 25 (left). Strength-


80 30 effective stress relation-
ships for Weald C’=v
A NEW RING SHEAR APPARATUS 313
The residual strength measured on undisturbed or shear zone samples from Arlington in
multiple reversal direct shear box tests (Table 7 and Fig. 24) c’~=O*~ lb/sq. in., +‘p= 105”, is
higher than that obtained from the ring shear test. The two series of tests on the undisturbed
clay were on samples from test pits 2 and 3, apparently with very similar Atterberg limits yet
widely different clay fractions.23 The ‘typical’ samples having the lower clay fraction show a
small but consistently higher peak and residual strength, as is to be expected. Similar tests
on clay from the shear zone in pit 2 show a lower peak strength, cl= 1.9 Ib/sq. in., I$‘= 13*3”,
and the same residual strength, c’=0.9 lb/sq. in., +‘= 10.5”. It is difficult to fit a curved
envelope to these three sets of residual points unless it is very sharply curved below ufn=5
lb/sq. in (see inset, Fig. 24), which seems improbable. Alternatively all these tests must be in
error. If c’ is assumed to be zero and 4’ remains at 105” then agreement with the ring shear
tests would be much better, but this is probably fortuitous in view of the much greater lack of
agreement in all other similar tests described in this Paper.
The ring shear test data on the Arlington sample are compared with triaxial compression
tests on oriented samples containing a slip surface in weathered Weald Clay from Sevenoaks
in Fig. 25. Samples from Pit 317H and Pit 300H again show comparable Atterberg limits but
different clay fractions and residual strengths. The Pit 317H samples with the higher clay
fraction show a reduction in $‘= of about 2”. However, the Atterberg limits and clay fraction
of the Arlington samples are in good agreement with the Pit 300H samples where the dif-
ference in residual strength is as much as 5”. The cohesion intercepts obtained in the triaxial
tests could be due to curvature of the envelope at normal stresses below about 15 lb/sq. in. or,
and this seems more likely, to membrane correction uncertainties at the lower stress levels.
The residual strength of Weald Clay from Arlington measured in the multistage ring shear
test is consistently lower than that obtained in multiple reversal direct shear box tests on clay
from the same site. The residual strength, as measured in triaxial tests on the slip surface, of
Weald Clay from Sevenoaks shows similar differences. This suggests that the inadequacy of
the direct shear box or triaxial test for determining residual strength also applies to this clay.

Studenterlunden Clay
A series of tests were carried out on a remoulded medium plastic clay from Studenterlunden
in Oslo (Norwegian Geotechnical Institute, 1968). Salt water of a similar concentration to the
pore fluid was added to bring the natural water content of the sample to the liquid limit, the
same salt solution being used in the water bath. The results are shown in Table 8 and Fig. 26.
The confining ring gap was 0.10-0.15 mm and the measured side-friction force was 21-38% of
the gross normal load-rather higher than the 10% measured by Garga (1970) at I.C. In
some cases the test was stopped overnight or at weekends, but this made no observable dif-
ference.
The ‘curve-hopping’ procedure developed by Schmertmann and Osterberg (1960) and
Schmertmann (1963) was followed in all tests. The net normal stress D’~was varied at about
3 mm displacement intervals, this interval being long enough to allow each stress-displacement
curve to be clearly established. Both the peak and residual strengths obtained from the six
tests lie very close to perfect straight lines through the origin, there being almost no scatter.
The peak strength $‘=31*6” is only l-6” higher than the residual. In tests 4/l and 5/l the
samples were over-consolidated at the start of shearing. However, both the peak and residual
strengths were indistinguishable from the remaining four normally-consolidated samples.
The second stage loading points obtained in tests l-6 by ‘curve-hopping’ all show excellent
agreement with the first stage points. The sixth and final loading stage of test 6, after 183 mm
displacement, gives an almost identical residual strength to the first stage. Also at the end of

23The difference in clay fraction was confirmed by repeated tests. The samples from pit 3 were atypical
of the Weald Clay from Arlington, being highly weathered and more fissured than usual.
Table 8. Results of ring shear tests at N.G.I. on Studenterlnnden Clay from Oslo
w.c.=43%, LL=41%, PL=21%, CF=38%. X-ray diffraction : quartz 22 % ; feldspar 17% ; illite 34 y0 ; chlorite 24 % ; hornblendes l-2 y0 ; organic 1%.
-
Test/stage O.C.R. Peak stresses Peak strength Residual stresses I Residual Shearing End of test
at - - strength rate,
start Normal Shear Normal Shear cm/min Displace- Elapsed
of
I a’,, ment, time,
fJnr
shear kg/cm2 kgGm2 kg/cm2 kg jc’me cm days

Remoulded
l/l 2.00 1.21 2.00 1.15 29.9”
l/2 1.74 1.08 1.74 1.01 30.1” 4.2 8.6
_ _-
2/l 1.00 0.59 30.5” 1.00 0.56 30.2”
212 0.87 0.54 31.6” 0.87 0.51 30.s” 3.7 7.6
_- -- _ _-
3/l 0.50 0.31 31.3” 0.50 0.30 30.50
312 0.42 0.27 32.5” 0.42 0.26 31.1” 4.2 8.6
_- 31.6’
4/l 1.oo 0.62 31.8” 1.00 0.58 30.3”
412 0.85 0.54 32.4’ 0.85 0.51 30.7” 30.00 0.00033 4.3 9.0
_- -- _ _-
5/l 0.50 0.31 31.8’ 0.50 0.29 30.1”
512 0.42 0.27 31.9” 0.42 0.26 30.9” 4.4 9.1
_- -- _ _-
6/l 3.00 1.81 31.1” 3.00 1.67 29.1”
612 2.55 1.56 31.4O 2.55 1.43 29.2”
- - 3.45 1.91 29.0”
6/3
614 - - 2.55 1.43 29.3O
615 - - 2,25 1.29 29.80
616 - - 3.00 1.66 29.0” 18.3 23.8
- - - I
A NEW RING SHEAR APPARATUS 315
Fig. 26 (right). Strength-
effective stress relation-
ships for remoulded Stu-
denterlunden Clay from -2
Oslo
P
E
; PEAK STRESSES
F
?I
‘.

B o Normally consolidated
.I?
v) . Over consolidated

c-’

I 2 3 4
q{ Normal effective stress, kg/cm’

I I 1 I I
4/2 denotes test 4, loadinq stoqe 2,etc

%2
<
_Y RESIDUAL STRESSES

r*
8
z
“,
;’

f
v,
Fig. 27 (below). Development
of residual strength of re-
moulded Cucaracha Shale
with increasing displace-
ment I 2 3 4
cq’ Normal effective stress, kg/cm’
0.5
Cucarocha Shale from Panama
Ring shear test at Imperial College
Remoulded sample
W.C.E48) LL=63, Pl=43, CF-48

0.4 - Nominal q’= 30.0 lb&in (2.11 kq/cm2)


Nominal shearing rote 0~00015in./min
Nominal 0.003in. confininq ring 90~

03

7
-,
Q-n

02

0.1 -

0
0.1
Displacement in inches
‘eat Residual
Drdincd ring shear test at M.I.T.
Reredimented:
w.c.=34,LL=60, PL=32, o-52 Herrmonn
Drained multiple reversal direct and
I4 shear box ,est 01 M.I.T.
Reredimented: Wolfskill
Peak strength envelope
r 0966)
W.C. - 3 b
Sheor CF-32
LL =60
rtrenqth Reredimcnted CF=52 PLE32 J
kqlcm’
d Multiple reversal direct iheor
Kellney
I.0 I box ‘smear’ test. Remoulded:
OCR’100 rc.=42, LL=59, PL=32, CF=36 1 (1967, 1970)
Droincd rinq shear test crt Harvard
Slaked and crushed:
w.=l25, LL-56, PL-42, CF-100 La Gatta
Best fit cur ved envelope Sloked: 0970)
I LL=49. PL=28. CF-42
0.5
Cut-plane mul;iple reversal
direct shear box test ot W.E.S. W.E.S.
Undirturbed: overoqe of 9 tests. (1970)
w.c.=19, LL=bS, PL=30, CF=27
Drained rinq shear test at I.C.
torqo
Remoulded:
2.0 2.5 3.0 w.c.=48, LL=63, PL=43, CF= 48 0970)
%‘ Normal effective ttresr, kq/cm’ .

4
I
7
Shear
strenqtt
kq/cm’
1

..
q’ Normal effective stress, kq/cm’
Fig. 28. Strength-effective stress relationships for Cucaracha Shale
A NEW RING SHEAR APPARATUS 317
test 6, before the final 15 mm displacement, the confining ring gap was increased from 0*1-O-15
mm up to 0.5 mm, during which time no change in the residual strength was observed. These
results on the Studenterlunden Clay, which has a relatively high residual strength, further
demonstrate the validity of multistage loading as a method of completely defining the residual
strength-effective stress relationship by testing a single sample only.

Cucaracha Shale
A single ring shear test was performed on a remoulded sample of Cucaracha Shale (Fig. 27).
The sample, which had apparently dried and decomposed owing to imperfect sealing and long
storage, was ground to pass a No. 36 B.S. sieve and remoulded up to a water content of 48%.
The value of 4 dropped from 28.0” at peak (c’ = 0) to 8.6” after a displacement of 3 in. with
a subsequent drop of 0.4” after a further displacement of 7 in. when a residual strength,
$‘,=8*2”, was reached. The data from this test are given in Table 9 and Fig. 28, together
with residual strength data on the Cucaracha Shale from four additional sources.
Tests at the Massachusetts Institute of Technology by Herrmann and Wolfskill (1966)
include ring shear tests using Hvorslev’s apparatus and also multiple reversal direct shear box
tests. The resedimented samples were prepared from material ground to pass a No. 40 U.S.
sieve, mixed with distilled water to a water content at least twice the liquid limit and con-
solidated under 5 kg/cm2. Samples from two sites were tested, but showed no significant
difference in strength. The residual strength determined from the M.I.T. ring shear tests
may be incorrect due to insufficient displacement and too rapid a shearing rate. The multiple
reversal direct shear box tests are also open to the objection that although as many as 40
reversals were imposed, excessive squeeze occurred. A markedly concave shear surface was
frequently observed after the test and the interpretation of the stress-displacement curves
becomes a matter of some conjecture. In spite of these qualifications, Herrmann and Wolf-
skill’s results are reasonably self-consistent and the average residual strength is similar to
Garga’s single test.
Multistage loading multiple reversal direct shear box tests on a ‘smear’ sample of remoulded
Cucaracha Shale by Kenney (1967, 1970) gave a residual strength of 6*1”, which is lower than
all other tests in the range a’, = 1-S kg/cm2. The 80 mm dia. sample was unconfined at
the edges and was sheared back and forth 2-2.5 mm either side of the centre position. Only a
limited number of reversals, five or six, were apparently imposed, the thickness being steadily
reduced due to squeezing.
A series of drained ring shear tests on remoulded samples at Harvard (La Gatta, 1970) gave
a significantly curved envelope with c’, =0.16 kg/cm2 and $‘r=6*2” for normal stresses of
4 kg/cm2 and above. Alternatively, assuming a straight line envelope through the origin,
$‘,=7*3”. At the lower stress levels the value of $‘= increases. At u’~= 1.0, 2.0 and 4-O
kg/cm2, #lr=9*7, 8.8 and 7*7”, respectively. The majority of La Gatta’s tests were on clay
which had been slaked four times and then crushed in a disc mill, but two tests were made on
uncrushed material. Although the clay fraction of the crushed clay was very much greater
than the uncrushed material, the residual strength was similar. However, the displacement
required to reach the residual state was greater for the uncrushed clay.24 The two over-
consolidated samples show an increase in remoulded peak strength with an increase in over-
consolidation ratio, as would be expected (Fig. 28). In three of the tests on the crushed clay
the normal stress was increased once a residual state had been achieved, i.e. multistage loading
was used. The residual strengths measured during the second loading stages are consistent
with those obtained during primary loading and with the curved envelope. At the end of the
test at 8 kg/cm2 normal stress the rate of displacement was first reduced by a factor of 10 and

a*This suggests that the large displacement required to define the residual strength may be associated
with the progressive breaking down of aggregates of clay particles.
0
Table 9. Results of residual strength tests on Cucaracha Shale: direct shear box and rim shear tests
z

Sample and test type Peak stresses Residual stresses Residual Initial 1Approx. End of test Reference
_______ strength water Shearing dis-
Normal
, Shear Normal Shear content, rate,, place- Dis-
e n, 7, a’“, (&O) % cm/mm ment, place-
kg/cm2 kg/cm2 kg/cm2 kg;c’m2 to &r, ment,
ClIl cm
---_ ~--

Test pit, west bank Cucaracha Reach 2.0 1.oso 2.0 0.284 8.1” 36 - 45 19
_.__~ -___ -~
Drained, multiple reversal direct 8.0 3.08 8.0 0.908 6.5’ 38 -
shear box test at M.I.T. - - ;:; 0.513 7.30 ;; 1
Resedimented: - - 0.285 8.10 33 17
LL=60, PL=32, CF=32 v~,~~ _, _’ 0~00008 z ______
Sample: 5.1 x 5.1 x 2.4 cm thick 4.0 j 1.960 4.0 0.526 7.5” 35 - 45 19
~I _______ ~-~ ________
16.0 5.45 16.0 1.795 ;:$Z 38 I
- 4.0 0.582 :; 1
Herrmann
- 2.0 0.315 8.9” 32 17
_.--‘-_ --- -- and
4.0 1.432 1 4.0 0.652 9.3” 34 - Wolfskill
Test pit, Cerro Escobar Slope 15 51
--,- -__~ ____- ~____ (1966)
Resedimented:
LL.=60, PL=32, CF=52
Drained, multiple reversal direct 2.0 ( 0.842 / 2.0 ( 0.346 / 9% 1 36 ( o’ooo12j -
shear box test at M.I.T. __- .___
Sample : 7.6 x 7.6 x 1.9 cm thick 7.50 37 - 30 17
--- --
Drained ring shear test at M.I.T. 7 0.06

Sample: o.d. = 11.4, i.d. =6.4, 1.9 cm


thick 10.10 19 29
-.
Drained, multiple reversal direct
shear box ‘smear’ test at N.G.I. - Kenney
Sample: 8 cm dia. x 0.2-0.3 cm thick (1967, ;
Remoulded : LL = 59, PL = 32, CF = 36 1970)
_. --- z
Hodges Hill, Hole CRW-15, Sample 13 1.0 0.410 1.0 / 0.195 ’ I 1 7 1
z
::::: i+i tr
Drained ring shear test 1.0 I 0.415 I 1.0 I-_I0.190 3 23 3
___~ LI
Sample: o.d.=7.1, i.d.=51,
0.2-0.3 cm thick U
Slaked 4 times and crushed 1.0 1 0.420 / I.0 j 0.151 / 8.6” j Ii:1 10 28 4
LL= 156, PL=42, CF= 100
2.0 / 0.660 126 / 1
:
- I -
4.0 0.570 8.1” 1 22
8.0 1.020 ( 7.3” / 133 / - 65 i 1>a Gatta
I !l”I lY!O
-1 -I- __ 0.0048 _ ~- (1970)
12.0 2.660 12.0 1.490 7.1” 115 2 50
- - 20.0 2.358 6.7” - 71 :
__ _ __~___
8.0 1.615 8.0 0.980 7.0°* 128 2 14 2
__ _
O.C.R. = 10 1.0 / 0.655 1 1.0 0.160 9.1” 116 5 15 2
~~
O.C.R.= 100 1.0 0.910 1.0 0.180 10.2O 117 10 20 3
__--- _. ---
Slaked 4 times 1.0 0.437 1 1.0 0.140 8.0” - 70 122 8 -I
LL=49, PL=28, CF=42 _. _
4.0 ( 1.564 1 4.0 0.575 8.2” - 20 34 4
__ _- _---- --
Drained ring shear test at I.C. 2.11 1.12 2.11 0.00038 8 28 51 (
Remoulded : LL = 63, PL = 43, CF = 48 o’305
I 8’2”
I 48
- -. _--.- _-
LL PL CF
Cut-plane, multiple reversal _-------
drained direct shear box 60 26 38 1
test t:; - :: 7
Undisturbed East _ _. _ ___~
Sample : Culebra 65 29 28 8.0 0.96 22 - 18 -
7.6 x 7.6 x 2.5 cm thick slide 4.0 0.48 ;:;I - 24 -
_ _. _~~
61 1 25 1 33 / E33; 0.96 6.8” 14 3
0.40 5.7” - ;: z
_. ---
West 83 35 24 8.0 0.80 5.7” 27 5 gater- 2
Culebra 4.0 0.40 5.7” - ;; I ways ;
slide ------ _- -p- Experi-
49 26 33 1.20 8.5” 16 0.00008 1 10 - ment z
2:; 0.56 8.0” - 20 - Station c
c/l
_. _ (1970)
Model 49 25 22 8.0 2 -
slope 4.0 - 85 -
_-_-__ _. _ -1

48 27 22 8.0 0.96 6.9" 16 2 10 -


4.0 0.44 6.3” - 20 -
-. I i ~~
102 42 27 8.0 1
4.0 - ;: 1
-_----___ _. ~___
52 31 13 8.0 1.08 18 6
4.0 0.58 -
- - -
* Rate of displacement at end of test: (a) reduced tenfold, no change in +‘r over 10 cm displacement, and then (b) increased ow z undredfold, Cs
no change in +‘= over 18 cm displacement. z
320 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

secondly increased by 100. The resultant effect on the residual strength was negligible, con-
firming that full drainage occurred and that time effects were small.
The peak strength of the slaked but uncrushed clay (La Gatta, 1970) agrees closely with
that of the resedimented clay (Herrmann and Wolfskill, 1966), whereas the slaked and crushed
clay shows a lower peak strength. Crushing has broken down the silt and fine sand sized
aggregates of clay particles, but although this reduces the peak strength, the residual strength
is unaffected.
The cut-plane multiple reversal direct shear box tests on undisturbed samples show
some scatter (Waterways Experiment Station, 1970). The intact specimens were pre-cut
with a bandsaw and the two halves of the shear box were adjusted to leave a l/16 in. gap. Up
to 25 reversals were imposed under an initial 8 kg/cm2 normal stress, the residual strength
generally being attained after the second or fifth forward travel and thereafter remaining
fairly constant. The samples were then unloaded to 4 kg/cm2 and the residual strength re-
established by imposing additional displacements. The nine tests give average residual
strengths of 7.0 and 7.3” at orn= 4 and 8 kg/cm2 respectively, which are in good agreement with
La Gatta’s tests. In principle, a cut-plane multiple reversal direct shear box test on a very
stiff material, such as the Cucaracha Shale, is likely to give a better estimate of $‘= than for
soft material, such as London Clay, since the two halves of the box can be well separated and
20 or 30 reversals can be imposed with less squeezing. Even so it is not certain whether the
scatter on the W.E.S. data is due to variations in the shale samples or to deficiencies in
apparatus and technique.
It should be emphasized that only in the series of tests on the Cucaracha Shale does the
residual strength measured by multiple reversal direct shear box tests agree with the ring shear
value.

DISCUSSION AND CONCLUSIONS


In any discussion of a new piece of testing equipment two questions must be asked. Do
the results represent a correct measurement of the relevant soil property? If so, what is their
significance?
The first question may be answered in two ways. Fig. 3 may be re-examined to see what
errors may enter into the observation and interpretation of the test results. It will be seen
from Fig. 3 that, when the table carrying the lower part of the soil sample is being rotated with
the upper confining rings lifted to avoid metal to metal contact, the whole shear load carried

Table 10. Relationships between residual strengths measured in the shear box and in the ring
shear apparatus

Clay type Residual strength s


in ring shear test,
T&S

a’,=5 am = 15 ~
lb/sq. in. lb/sq. in.

Blue London 0.83 2.48


Brown
London 0.91 2.37
Weald 1.35 3.25 1.83 3.68 1.36 1.13

a’,=2 ofn=4 c/,=2 a’,=4 a’,=2 IT’,=4


kg/cm2 kg/cm2 kg/cm2 kg/cm2 kg/cm2 kg/cm2

Cucaracha
shale 0.31 0.60 0.31 0.52 1.00 0.87
I -
A NEW RING SHEAR APPARATUS 321
by the clay in the plane of relative movement is transmitted to the torque arm and yoke
carrying the upper platen and upper confining rings, respectively. The only errors that can
arise (other than in the calibration of the proving rings) are due to friction in the ball bushings
guiding the torque arm and in the two swivels.
Since the outer housing of the lower ball bushing accommodating the main shaft rotates
with the table, any tangential frictional force would tend to increase the observed torque and
thus increase the apparent value of the residual strength. The main swivel, which transmits
the vertical load from the lever system to the torque arm and thence to the upper platen, is
only subject to the small angular movement resulting from the compression of the proving
rings. In addition, it consists of two separate ball thrust races in a housing (Fig. 6) which can
be rotated through a substantial angle in either direction while under the vertical load to
eliminate any possibility of a build-up of torque.
The vertical load on the upper swivel arises only from the side friction of the sample and
upper platen on the upper confining rings and is only a few per cent of the load on the main
swivel. The angular movement is likewise small. The provision of similar ball thrust races
is considered sufficient to substantially eliminate any torque, which, in contrast to that of the
main shaft, would tend to reduce the apparent value of the residual strength.
The sum of these three possible sources of frictional error is thus considered to have an
entirely insignificant effect on the measurement of torque.
The main shaft also transmits the normal load to the upper platen. Since the ball bushing
guiding it is rotating with the table carrying the sample, any possibility of a vertical component
of friction will be destroyed during the shear stage of the test by the relatively much greater
circumferential movement of the ball bushing. Friction, if any, in the upper ball bushing
merely affects the distribution between the top and the sides of the specimen of the net vertical
force,25 which is the difference between the load applied through the lever system and the load
observed on the vertical proving ring below the screw controlling the gap between the rings.
The only other possible source of error in the normal load is at the contact between the
torque arm and the proving rings. Here small steel wheels are used, bearing on hardened
plates set in the torque arm, so that negligible vertical friction results.
The errors in vertical loading are therefore considered to lie entirely within acceptable
limits.
Gross errors in the observations of torque and vertical load may thus be ruled out, but the
question of interpretation remains. This has been dealt with in detail in the section on cali-
bration and evaluation of the results. It is apparent from this Section that, for the relatively
low ratio of breadth to mean diameter of 1: 5, the calculated value of the residual angle is not
significantly affected by even the most extreme assumption made about the radial distribution
of normal stress. The assumption that rupture is occurring on a surface approximating to the
plane of relative motion is confirmed by direct observation (Figs 13 and 14).
A strong case can therefore be made for the view that the results represent a correct
measurement of the residual strength. This case is further supported by La Gatta’s indepen-
dent observations in the Harvard apparatus, which differs in both scale and mechanical detail
but gives an average value agreeing to within 0.1” in the case of the blue London Clay from
Wraysbury (Fig. 18 and Table 2) on which a full series of tests was carried out in each
apparatus.
The results, if their correctness is accepted, are significant in a number of respects. They
show very clearly that in general the multiple reversal direct shear box test gives a result which
in the case of clays differs very substantially from the true residual strength. This is not to
say that the multiple reversal direct shear box test is not of value in providing an empirical
correlation with field observations. However, the values given in Table 10 show such a wide
25It is this net vertical force that is carried by the annulus of clay in the plane of relative movement
when the gap between the rings is open.
322 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

divergence between the strengths given by the ring shear apparatus and by the multiple rever-
sal direct shear box26 tests that both cannot claim to be an acceptable way of measuring the
same parameter.
The fundamental difference between the two tests lies in the failure of the multiple reversal
direct shear box test to simulate the field condition of a large relative displacement uninter-
rupted by changes in direction. This difference is reflected in the appearance of the slip sur-
face at the stage of the test at which the residual strength is defined. Fig. 20 of Skempton
(1964) shows the slip surface formed by a multiple reversal direct shear box test and this may
be compared with Fig. 14 of this Paper, in which the slip surface is clearly smoother and more
highly polished.
Tests in the triaxial apparatus on 1.5 in. dia. specimens and in the shear box (for example,
Tchalenko, 1967) indicate that displacements of the order of O-3 in. may give rise to a contin-
uous rupture surface. The results of the ring shear test on blue London Clay presented in
Fig. 12 indicate that in this case ‘rupture surfaces’ can exist on which the available angle of
shearing resistance varies from about 17” for small displacements to 9.4“ at large displacements.
Which value is mobilized in any particular engineering problem is clearly a matter of
crucial importance. The two tests on samples cut from a slip surface in the field at Wrays-
bury on which a rapid undrained displacement of about 5 ft had taken place gave an average
value of Cfr of 9.2” (Table 3). This is indistinguishable from the ring shear value.
However, the testing of samples cut from a slip surface and transferred to a testing
machine is subject to several hazards. As described earlier, a block of clay from Wraysbury
was sheared undrained to a very large displacement in a torsion shear apparatus, and a dia-
gonal sample was cut to include the slip surface. When tested under drained conditions in
the triaxial apparatus this gave an apparent value of $‘= = 10~5”(Table 3), about lo higher than
the ring shear value. The sample, and thus the rupture surface, was subject to a high negative
pore pressure when the clay was unconfined at the sample preparation stage. This may affect
the orientation of the clay particles on or close to the slip surface.
The multistage ring shear test on the brown London Clay from Walthamstow illustrates a
further effect (Fig. 19). Although at each stage a surface had been formed which defined the
residual strength corresponding to a particular normal stress, a further displacement of l-2 in.
was required to re-establish the residual state for the next loading stage if this was associated
with a reduction in effective stress. A knowledge of previous stress history and the avoid-
ance of surface drying may thus be essential to the correct evaluation of $‘= in laboratory tests
on natural shear surfaces or artificially prepared planes.
The problem of correlating laboratory values of 4’r with field observations on fullscale slips
raises many issues which are outside the scope of this Paper. The value of #‘r estimated from
field observations is very sensitive to the accuracy of the pore-pressure observations and to
their relevance in terms of the positioning of the piezometers, particularly when the depth to
the slip surface is small. In many cases the average normal stress acting on the failure
plane in the field is low; often less than 10 lb/sq. in. according to De Lory (1957) and James
(1970). Thus the incorrect high values of $‘r determined for the brown London Clay and for
the Weald Clay from multiple reversal direct shear box tests with up to six reversals at normal
stress levels generally well in excess of 10 lb/sq. in. are fortuitously similar to the more correct
values of $‘r determined in the ring shear tests at the more appropriate lower stress levels.
This may help to explain the correlation between the field and laboratory multiple reversal
direct shear box test residual strengths in spite of the latter’s inaccuracy.
26The plots of the mobilised shear stress (as a ratio of ~10’~) against displacement for the multiple reversal
direct shear box tests in Figs 15 and 16 show that, after an early minimum, the values of T/U’,, are rising as
each shear stage terminates except for the first stage which defines the peak value. Neither the minimum
value nor the value at the end of the reversal stages appears to show a consistent relationship with the values
given by the ring shear apparatus. In the M.I.T. tests the minimum value has been used and for the
Cucaracha Shale this is in reasonable agreement with the ring shear. Other investigators have used the
terminal value.
A NEW RING SHEAR APPARATUS 323

60
Brittleness
lndcx

1.
%

0
0 IO 20 40 50
Effective normal I~TCSS, rn’ Ib/sq in.

Fig. 29. Variation in brittleness index IB with effective normal stress for various soils

However, the direct relevance of the residual strength to most slides in cuts and excava-
tions, even in stiff-fissured clays, has been questioned. Skempton (1969, 1970) has postulated
that in ‘first-time’ slides the failure may be explained by the uniform mobilization of a ‘fully-
softened’ strength, which corresponds closely to the peak strength of normally-consolidated
remoulded samples. The residual value is, on this basis, only relevant to the re-initiation of
movement on surfaces resulting from earlier slides or tectonic movement.
An alternative view (Bishop, 1971) is that first-time slides are examples of progressive
failure. At the moment of limiting equilibrium the strength will vary along the failure surface
from the peak value down to a value at or close to the residual at the point where local over-
stress initiates failure. In this case the value of the residual strength is relevant, but an exact
analysis of stability requires a knowledge of the extent of the reduction from peak to residual
along the failure surface. It is of interest to note that Hvorslev (1939) had progressive failure
in mind in his explanation of the need for measuring strength at very large displacements.
Even the analysis of the re-initiation of slides on old surfaces may need to be viewed with
some caution if the effective normal stress at which the original surface was formed was higher
than that existing when renewed movement took place. As indicated earlier, there is evidence
to show that appreciable further displacements may be required before a new residual state is
established. The field value of c$‘~from a re-initiated slide would in this case tend to be greater
than the laboratory value of the residual strength.
In conclusion it is of interest to consider whether the residual strength of clay soils of high
plasticity is inherently more relevant to the solution of stability problems occurring in engin-
eering practice than that of low plasticity clays and sands. In this connexion it is convenient
to consider the brittleness of the various clays and sands, using the brittleness index I,,
defined by Bishop (1967) :

where 7f denotes the shear stress at failure (peak),


and 7, denotes the residual shear stress, the normal effective stress being unchanged.
In Fig. 29 the variation in brittleness index with effective normal stress is plotted for
324 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

various soils. This Figure illustrates the relative importance of three factors which contribute
to the brittleness of a soil. These are:
(a) the dilatancy accompanying failure,
(b) the reorientation of clay particles adjacent to the slip surface,
(c) cementation bonds between particles and particle groups.
Taylor (1948) and Bishop (1950) demonstrated that in cohesionless sands the difference in
strength between the peak and the residual state (which can in this case be identified with the
ultimate or critical state) is directly related to the rate of dilatancy at failure. This difference
can be very significant. While in Taylor’s example I, was only 26%, denser samples showing
a higher rate of dilatancy can give I, values of 3040% in the triaxial test (curve 1, Fig. 29)
and 4O-5OO/oin plane strain (curve la, Fig. 29), the values decreasing with increase of effective
normal stress.
These values of I, are substantially below those of clays of high plasticity, but are asso-
ciated with only relatively small strains before the residual state is achieved.27 It is of interest,
therefore, to note that considerable error has been found to result if progressive failure is
ignored in the analysis of the results of model tests in plane strain in the low stress range in
dense sand (Rowe and Peaker, 1965; Rowe, 1969).
The second factor, the reorientation of the clay particles in the zone of failure, was referred
to by Hvorslev (1960) and was directly related to the residual state by Skempton (1964). The
magnitude of this factor can be isolated by comparing the residual value with the peak value
of +‘r obtained for remoulded samples either from constant volume tests or from tests with
negligible dilatancy. Consolidated-undrained tests with pore pressure measurement gave
values of dfr of 21” in the low stress range falling to 16-l” at a normal effective stress of
approximately 600 lb/sq. in. (Bishop et al., 1965) on samples of blue London Clay from
Ashford Common consolidated from a slurry. The peak value from the one remoulded sample
of Wraysbury Clay tested in the ring shear apparatus was 19-l” at a normal effective stress
of 26.9 lb/sq. in., the dilatancy being very small. An average value of 20.0” may
reasonably be assumed over the pressure range illustrated in Fig. 29.
Since the value of $‘r for the blue London Clay from Wraysbury is independent of normal
stress in the low pressure range, this leads to a value of the brittleness index of 55% throughout
this range for the remoulded clay (curve 2, Fig. 29). AS cementation bonds have been largely
destroyed by remoulding, and dilatancy is absent, the brittleness of remoulded clay must be
attributed wholly to the reorientation of the platey clay particles.
It is of interest to note that the value of Ia is greater for remoulded blue London Clay than
for dense sand in plane strain, yet no problem of progressive failure has been reported in re-
moulded clay. Reorientation of the clay particles may therefore be the factor in residual
strength which requires very large strains or relative displacements for its full development.
This is supported by the observation that the shape of the post-peak stress-displacement curve
for remoulded clay does not differ substantially from that for undisturbed clay (in the case of
blue London Clay).
The existence of the third factor, cementation bonds, as a significant component of the peak
strength of the blue London Clay is demonstrated by the substantial tensile strength shown by
intact samples in drained tension tests (Bishop and Garga, 1969). The IB curves for tests on
undisturbed samples of weathered blue London Clay (curve 3, Fig. 29), unweathered blue
London Clay (curve 4) and brown London Clay (curve 5) all have two features in common.
The I, values are high, being above 70% for the stress range below 30 lb/sq. in. in which most
27Tests in plane strain carried out at I.C. (Cornforth, 1961) suggest that the residual state may be
reached in sands at axial compressions as small as 10%. However, tests by Green (1969) suggest that in
brittle materials one effect of end restraint on the test results may be to induce complete zone failure soon
after the peak strength has been mobilized. In the simple shear test a close approximation to the residual
state is reached at a shear strain of the order of 1. However, the enforcement of uniform boundary displace-
ments does not completely inhibit the tendency to zone failure (Roscoe, 1967).
A NEW RING SHEAR APPARATUS 325
reported landslides in London Clay are found to lie. Secondly, the I, values rise rapidly at
very low stresses and all become lOOo/o at zero normal stress. This reflects the significant
cohesion intercept associated with the failure envelope for the undisturbed samples and the
absence of such an intercept in the residual state. The existence of this intercept in undis-
turbed samples is generally inferred from tests in which the effective normal stress is seldom
below 5 lb/sq. in. (Figs 17 and 21). Only a few tests have been carried out in the triaxial
apparatus 28 with the minor effective principal stress u’~ eq ual to almost zero (Bishop et al.,
1965), or negative (Bishop and Garga, 1969) but these provide strong evidence of the exis-
tence of cohesion at zero effective normal stress in relatively intact samples. While lower
values are to be expected in the weathered mass, zero cohesion would require a very marked
curvature in the failure envelope near the origin, which would have only a local effect on the
I, curves. Recent drained tests by Marsland (1971) on samples from the upper zone of the
blue London Clay from Wraysbury indicate that this may indeed be observed if large
samples of the fissured clay are tested. Marsland’s results suggest a value of I, of 70% as
zero normal stress is approached.
The relative importance of these factors depends on the stress level under consideration.
This may be illustrated most clearly by the test results presented in Fig. 17, which permit a
direct comparison to be made of the strength-effective stress relationship for the undis-
turbed peak, the remoulded peak and the residual states. The difference between the re-
moulded peak value and the undisturbed peak value may be considered to be due to the
destruction of cohesive bonds and the elimination of dilatancy.2g The difference between the
remoulded peak and the residual strength may be attributed to reorientation.
At a normal effective stress of 5 lb/sq. in., the loss in drained strength due to the destruction
of cohesive bonds and dilatancy is 73% of the peak value. The further loss due to orientation
is 14%. At a normal effective stress of 50 lb/sq. in. the losses are 25% and 39%, respectively.30
This suggests that in the low stress range in which the majority of examples of progressive
failure are to be found (Skempton, 1964, 1970) the destruction of cohesive bonds plays a more
significant role than the reorientation of the clay particles. At depths below ground surface
to which the higher stress ranges would be relevant, the magnitude of the cohesion intercept
tends to be much larger (Bishop et al., 1965), and as a consequence cohesion may tend to be the
major factor in progressive failure for earth movements of all magnitudes.
Soils such as the boulder clay at Selset (curve 6, Fig. 29), in which progressive failure plays
only a small part (Skempton, 1964), have low values of I, even in the relatively low effective
stress range.

ACKNOWLEDGEMENTS
The design of the new ring shear apparatus was carried out jointly by the Norwegian
Geotechnical Institute and Imperial College. The details were prepared by Mr K. Sundre of
N.G.I. and the apparatus was constructed in Norway.
Financial assistance for the equipment at Imperial College was provided by the National
Environmental Research Council (N.E.R.C.) in a grant to Professor A. W. Skempton. Tests
at I.C. on the blue and brown London Clay, the Weald Clay and the Cucaracha Shale were
performed by one of the Authors, Dr V. K. Garga, with the financial assistance, in part, of the
Construction Industry Research and Information Association (C.I.R.I.A.).
The Authors are grateful to the Metropolitan Water Board and John Laing Construction

a* The shear box test is in principle unsuitable for carrying out direct shear tests at low effective stresses
on soils showing substantial cohesion (Roscoe, 1953).
29 The role of dilatancy in strongly bonded soils is obscure, since both drained tests (with volume change)
and undrained tests (constant volume) give the same failure envelope (Bishop et al.. 1965). Dilatancy was
not very marked in the present series of tests (Figs 12 and 19).
30These values add up to the I, values presented in Fig. 29, curve 3.
326 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

Ltd for facilities for obtaining samples of the blue London Clay from Wraysbury. The tests
by Dr D. J. Petley on the brown London Clay from Walthamstow are published by kind per-
mission of the Borough Architect for the London Borough of Waltham Forest, who also pro-
vided facilities for obtaining the ring shear sample. The Authors are also grateful to the
Eastbourne Waterworks Company and Binnie & Partners for permission to quote test results
obtained by Dr D. J. Petley on the Weald Clay from Arlington Reservoir and for supplying
the ring shear sample.
One of the Authors, Dr J. D. Brown, was responsible for the tests at N.G.I. on the Studen-
terlunden Clay from Oslo, which were carried out by Mr L. Ruud.
Thanks are due to the Waterways Experiment Station for permission to use the data on the
Cucaracha Shale and for the sampIe tested at I.C. provided by Professor L. A. Wolfskill of
M.I.T. and to Professor T. C. Kenney of the University of Toronto for additional unpublished
test data on the Cucaracha Shale.
The Authors are grateful to Dr M. J. Hvorslev of the Waterways Experiment Station and
Mr A. D. M. Penman of the Building Research Station for an early discussion of the design.
Thanks are also due to Professor A. Casagrande and Dr D. La Gatta of Harvard for testing the
blue London Clay from Wraysbury and for helpful discussions.
Mr E. W. Harris has given valuable assistance in the preparation of figures for publication.

REFERENCES
AGARWAL, K. B. (1967). The influence of size and orientation of sample on the undrained strength of London
Clay. Ph.D. thesis, University of London.
AMERICAN SOCIETY OF CIVIL ENGINEERS(1917). Progress report of the special committee to codify present
practice on the bearing value of soils for foundations, Appendix A. PYOC.Am. Sot. Civ. Engrs, Papers
and discussions, 1174-l 191.
BISHOP, A. W. (1950). In discussion on: The measurement of the shear strength of soils. G&otechn@e 2,
No. 2, 113-116.
BISHOP, A. W. (1967). Discussion, Session 2. Shear strength of stiff clay. Proc. Geotech. Conf., Oslo
2, 142-150.
BISHOP, A. W. (197 1). The influence of progressive failure on the method of stability analysis. Gbotechnipue
21, No. 2, 168-172.
BISHOP, A. W. & GARGA, V. K. (1969). Drained tension tests on London Clay. Gtolechnique 19, No. 2,
309-313.
BISHOP, A. W., WEBB, D. L. & LEWIN, P. I. (1965). Undisturbed samples of London Clay from the Ashford
Common Shaft: Strength-effective stress relationships. Gdotechnique 15, No. 1, 1-31.
BLIGHT, G. E. (1967). Observations on the shear testing of indurated fissured clays. Proc. Geotech. Conf.,
Oslo 1, 92-102.
CARR, J. F. & WALKER, D. M. (1968). An annular shear cell for granular materials. Powder Technod. 1,
369-373.
CHANDLER, R. J. (1966). The measurement of residual strength in triaxial compression. Gkotechnipue 16,
No. 3, 181-186.
COOLING,L. F. & SMITH, D. B. (1935). The shearing resistance of soils. JnZ Instn Civ. Engrs 3.333-343.
COOLING, L. F. & SMITH, D. B. (1936). The shearing resistance of soils. Proc. 1st 1st. Conf. Soil Mech.,
Harvard 1, 3741.
CORNFORTH,D. H. (1961). Plane strain failure characteristics of a saturated sand. Ph.D. thesis, University
of London.
DE LORY, F. A. (1957). Long-term stability of slopes in over-consolidated clays. Ph.D. thesis, University of
London.
FOOKES, P. G. & PARRISH, D. (1968). Observations on small-scale discontinuities in the London CIay and
their relationships to regional geology. Q. Jnl Engng Geol. 1, 217-240.
FRANZIUS,O., STRECK, A. & HINDERKS, A. (1929). The Experiment Institute for Hydraulics and Founda-
tions at the Technical University at Hanover. Hydraulic laboratory practice 599-601. American
Society of Mechanical Engineers.
GARGA, V. K. (1970). Residual shear strength under large strailzs and the effect of sample size on the consolida-
tion of jissured clay. Ph.D. thesis, University of London.
GHANI, M. A. (1966). Determination of shear stress of a fine cloddy soil with a ‘guarded’ shear head.
PYOC. 2nd Int. Conf. Int. Sot. for Terrain-Vehicle Systems 279-310. Toronto: University of Toronto
Press.
GIBSON, R. E. (1951). An investigation of the fundamental shear strength characteristics of clays. Ph.D.
thesis. Universitv of London.
GREEN, G: E. (1969): ~Strength and compressibility of granular materials under generalised shain conditions.
2 volumes. Ph.D. thesis. University of London.
A NEW RING SHEAR APPARATUS 327
GRUNER, H. E. & HAEFELI, R. (1934). Beitrag zur Untersuchung des physikalischen und statischen Ver-
haltens Roharenter Bidden [Contribution to the investigation of the physical and static behaviour of
cohesive soils]. Schweiz. Bauztg 103, 171-174, 185-188.
HAEF~Y~~~~R~.(1938). Mechanische Eigenschaften von Lockergesteinen. Schreeiz. Bauztg 111, 299-303,

HERRMANN, ‘H. G. & WOLFSKILL, L. A. (1966). Residual shear strength of weak shales. Technical Refiort
No. 3-699, Engineering properties of nuclear craters, Report 5. Massachusetts: Massachusetts Institute of
Technology.
HUTCHINSON,J. H. (1969). Private communication.
HVORSLEV, M. J. (1937). Uber die Festigkeitseigenschaften gestcirter bindiger Bdden [On the strength proper-
ties of remoulded cohesive soils] 159 pp. Thesis, Published by Danmarks Naturvidenskabelige
Samfund, Ingeniorvidenskabelige Skrifter, Series A, No. 45, Copenhagen. Translated June 1969 by
USAE Waterways Experiment Station, Vicksburg, Mississippi.
HVORSLEV, M. J. (1939). Torsion shear tests and their place in the determination of the shearing resistance
of soils. Proc. Am. Sot. Test. Mater. 39, 999-1022.
HVORSLEV, M. J, (1960). Physical components of shear strength of saturated clays. Proc. Res. Conf. Shear
Strength Cohesive Soils, Boulder 169-273. American Society of Civil Engineers.
HVORSLEV, M. J. & KAUFMAN, R. I. (1952). Torsion shear apparatus and testing procedures, Bull, No. 38.
Vicksburg, Mississippi: USAE Waterways Experiment Station.
JAMES, P. M. (1970). Time effects and progressive failure in slopes. Ph.D. thesis, University of London.
KENNEY, T. C. (1967). The influence of mineral composition on the residual strength of natural soils.
Proc. Geotech. Conf., Oslo 1, 123-129.
KENNEY, T. C. (1970). Private communication.
LA GATTA, D. P. (1970). Residual strength of clays and clay-shales by rotation shear tests. Harvard Soil
Mechanics Series No. 86, Cambridge, Massachusetts.
LANGER, C. (1938). Proprietes mircaniques et physiques des terrains coherents. Compte rendu des recherches
effect&es durant 1’Annk 1938 19-27. Paris: Laboratoires du Batiment et des Travaux Publics.
LA ROCHELLE, P. (1960). The short-term stability of slo$es in London Clay. Ph.D. thesis, University of
London.
LA ROCHELLE, P. (1967). Membrane, drain and area correction in triaxial tests on soil samples failing along
a single shear plane. Proc. 3rd Pan. Am. Conf. Soil Mech. 1, 273-291.
MARSLANE, A. (1971). The shear strength of stiff -fissured clay. Proc. Roscoe Memorial Symp., Cambridge
(in press).
NORWEGIAN GEOTECHNICALINSTITUTE (1968). Torsion ring shear tests on remoulded Studenterlunden Clay.
Internal Report F346. Oslo: Norwegian Geotechnical Institute.
NOVOSAD, J. (1964). Studies on granular materials II. Apparatus for measuring the dynamic angle of
internal and external friction of granular materials. Colin Czech. Chem. Commun. Engl. Edn 29,
2697-2701.
NOWACKI, E. H. F. (1967). Anvendelse av ringskjaerapparat for studier av omrorte leirers egenskaper ued store
deformasjoner. Internal Report F277. Oslo: Norwegian Geotechnical Institute.
PAYNE, P. C. J. & FOUNTAINE, E. R. (1952). A field method of measuring the shear strength of soils.
Jnl Soil Sci. 3, 136-144.
PENMAN, A. D. M. (1953). Shear characteristics of a saturated silt, measured in triaxial compression.
Gdotechnique 3, No. 8. 312328.
PETLEY, D. J. (1966). The shear strength of soils at large strains. Ph.D. thesis, University of London.
PETLEY, D. J. (1969). Unpublished tests at Imperial College.
ROSCOE, K. H. (1953). An apparatus for the application of simple shear to soil samples. Proc. 3rd Int. Conf.
Soil Mech., Zurich 1, 186-191.
Ro~cl~&~~H. (1967). Discussion on shear strength of soil other than clay. Proc. Geotech. Conf., Oslo 2,

ROWE, P. W. (1969). Progressive failure and the strength of a sand mass. Proc. 7th Int. Conf. Soil Mech.,
Mexico 1, 341-349.
ROWE, P. W. & PEAKER, K. (1965). Passive earth pressure measurements. Geotechnique 15, No. 1, 57-78.
SCARLETT,B. &TODD, A. C. (1968). A split ring annular shear cell for the determination of the shear strength
of a powder. Jnl Scient. Instrum. Series 2, 1, 655-656.
SCHMERTMANN,J. H. & OSTERBERG. J. 0. (1960). An experimental study of the development of cohesion
and friction with axial strain in saturated cohesive soils. Proc. Res. Conf. Shear Strength Cohesive Soils,
648-694. Boulder: American Society of Civil Engineers.
SCHMERTMANN,J. H. (1963). Generalising and measuring the Hvorslev effective components of shear
resistance. Laboratory shear testing of soils. STP No. 361, 147-157. Ottawa: American Society for
Testing Materials.
SEMB;~~;~;; P. & RAMIREZ, A. L. (1969). Measurement of residual strength of clay with a rotating shear
Proc. 7th Int. Conf. Soal Mech., Mexico 3, 528-529.
SKEMPTON,A. W. (1964). Fourth Rankine Lecture. Long-term stability of clay slopes. Gdotechnique 14,
No. 2, 77-102.
SKEMPTON,A. W. (1969). Discussion on main session 5. Proc. 7th Znt. Conf. Soil Mech., Mexico 3,401-402.
SKEMPTON,A. W. (1970). First-time slides in over-consolidated clays. Geotechnique 20, No. 3, 320-324.
SKEMPTON,A. W. & HUTCHINSON,J. N. (1969). Stability of natural slopes and embankment foundations.
Proc. 7th Int. Conf. Soil Mech., Mexico. State-of-the-Art, 291-340.
328 A. W. BISHOP, G. E. GREEN, V. K. GARGA, A. ANDRESEN AND J. D. BROWN

SKEMPTON, A. W. & PETLEY, D. J. (1967). The strength along structural discontinuities


in stiff clays.
Proc. Geotech. Conf., Oslo 2. 29-46.
SKEMPTON, A. W., SCHUSTER, R. L. & PETLEY, D. J. (1969). J oints and fissures in the London Clay at
Wraysbury and Edgware. Gkotechnique 19, No. 2, 205-217.
STRECK, 0. (1928). Fortschritte auf dem Gebiet des Baugrundforschung [Progress made in the field of
foundation soil research]. Zentbl. Bauvew., No. 9.
TAYLOR, D. (1948). Fundamentals of soil mechanics. New York: Wiley.
TCHALENKO, J. S. (1967). The infEuence of shear and consolidation on the microscopic structure of some clays.
Ph.D. thesis, University of London.
TIEDEMANN, B. (1937). Uber die Schubfestigkeit bindiger BBden. Bautechnik 15, Nos 30 and 33,400-403,
433-435.
WARD, W. H., MARSLAND, A. & SAMUEL% S. G. (1965). Properties of the London Clay at the Ashford Com-
mon Shaft: In-situ and undrained strength tests. Gdotechnique 15, No. 4, 321-344.
WATERWAYS EXPERIMENT STATION (1970). Private communication.
WEBB, D. L. (1966). The mechanical properties of undisturbed samples of London Clay and Pierre Shale.
Ph.D. thesis, University of London.

APPENDIX: ANALYSIS OF THE RESULTS ASSUMING A NON-UNIFORM STRESS DISTRIBU-


TION ON THE FAILURE PLANE
The derivation of the equations given in Table 1 is outlined in the following steps.
The equation for the non-uniform stress distribution u’~ at any radius r is defined in terms of the maximum
stress P and the inner and outer radii +‘r and v2, respectively, where rr < Y<ye, i.e. :
06 = f(P, Y, Vl, r2) . . . . . . . . . . (‘1
Then, normal load :
w= ‘2 o’,Bnr dv . . . . . . . . f (2)
f ‘1
from which an expression for P in terms of W, rl and r2 is obtained. This is substituted in equation (1)
giving u’,, in terms of W, r, rl and 12.
Assuming 7 = fYn tan f$’ . . . . . . . . . . * (3)
then the moment

ii/l = ‘=2nr@ dr . . . . . . . . . . (4)


s ‘1
Substituting equation (3) in equation (4) gives an expression for tan 4’ in terms of M, W, rl and r,.
For example:

Case : Parabolic I
P(r, - Y)S
Here for rr<r<ve
O’xl= (ra

therefore load W = & .:’ (v~-Y)~ dr


s

= $(rz--u,)(v,f3r1)

~W(V~-Y)~
therefore a6 =
7@2-~1)3(~2f3Y1)
therefore 7 = A(r,--r)a
6W tan 4
where A=
7+a--Y1)3(+,2+3rr)

therefore moment Af = 2aA ,:’ (Y~-~)~Y~ dr


s

= g (Y~-Y~)~(Y~~+~~~Y~+~Y~~)

2 W (rza + 3r,r, + 6rr2)


=- tan 4’
5 (r2 + 3r,)

tan 4’ = LM (r2 + 3r1)


so that
2 W (rz2 + 3r,r*+ 6rx2)
which is the required relationship for +’ in terms of the normal load, the moment and the sample geometry.

You might also like