You are on page 1of 8

NOTES ON THE DESIGN OF THE LIQUID LIMIT DEVICE

by
PROFESSOR ARTHUR CASAGRANDE

SYp\‘OPSIS
From experience during the 30 years since he D’aprl-s l’expkrience acquise au tours des 30 ans
built the first liquid limit device, and from his depuis qu’il construisit le premier appareil & limite
observations on the use and abuse of this device in de liquidit&, et d’apr&s ses observations sur l’emploi
the Author concludes that et l’abus d’emploi de cet appareil dans de nombreux
many laboratories,
laboratoires, I’auteur conclut que la rkp6tition pr&
accurate duplication of results requires improved
cise des rksultats demande une normalisation
standardization of several details in its design. amkliorke de plusieurs details dans sa conception.
The design shown in Figs 1 and 2 is recommended Le modkle rep&sent6 figures 1 et 2 est recom-
as a standard. It maintains the basic dimensions mandk comme modble normal. 11 garde les dimen-
and hardness of base which have been accepted by sions principales et la duet6 de base acceptCes par
tacit agreement and have been in use in many accord tacite et utiliskes dans de nombreux pays
countries for about a quarter of a century. depuis un quart de sikle.
In discussing the complications arising from the En discutant des complications survenant de
l’emploi d’une durct6 de base substantiellement
use of a substantially different hardness of base in
diffkrente de la norme britannique, l’auteur suggkre
the British Standard, the Author suggests as a comme solution possible le remplacement de
possible solution replacement of the liquid limit l’kpreuve de limite de liquiditk par une Bpreuve
test by a more rational test based on the measure- plus rationnelle baske sur la mesure de la resistance
ment of shear strength. au cisaillement.

INTRODUCTION
Early in this century, the Swedish soil scientist, A. Atterberg, carried out investigations
on the plasticity of fine-grainecl soils, which led him to the realization that plasticity is a
two-dimensional property that is best defined by the upper and lower limits of the plastic
range. He suggested two simple tests to define these limits, and named them “ Fliessgrenze,”
the liquid limit, and “ Ausrollgrenze,” the plastic limit (Atterberg, 1911).1 They are often
referred to as the plasticity tests, and also the Atterberg limits.
Professor Terzaghi, in his early investigations, recognizecl the importance of these plasticity
tests in soil mechanics. One of the first extensive practical applications of these tests was
carried out by the U.S. Bureau of Public Roads under the direction of Terzaghi (1926).
Terzaghi was aware that the liquid limit test as proposed by Atterberg was very difficult
to duplicate by different operators. In 1927 he suggested that the Author should try to
develop a mechanical device which would eliminate as far as possible the uncertainties of the
hand method. After a careful study of Atterberg’s publications the Author concluded that
a preceding attempt at mechanization had failed because it departed too far from the principle
of the original test. Therefore, the Author tried to adhere as closely as possible to the simple
hand test invented and used by Atterberg. This was readily achieved by means of a cam
that raised a cup to which was attached a cam follower. The difficulty was to find a com-
bination of materials for the base and the cup, and a method for controlling the dimensions
of the groove, which would permit duplication of results with reasonable accuracy. This
development work is summarizecl by Casagrande (1932).

THE ORIGINAL STANDARD


The following details were found to be critical and had to be standarclizecl :-
(u) Material of the base. Wood and various metals were found to be unsuitable.
Finally, the hardest type of hard rubber (ebonite), of a kind that could be machined and
1 The references are given on page 91.
84
NOTES ON DESIGN OF LIQUID LIMIT DEVICE 85

which 30 years ago was extensively manufactured for electrical equipment, was chosen.
The hardness of this material as obtained from different sources, appeared not to vary to
an objectionable degree. Later, as such devices were built in other countries, several
comparisons were made of types of hard rubber used abroad with that used in the United
States of America, with satisfactory results. For such comparisons parallel tests were
made on identical soil rather than relying on a hardness test. Unfortunately, no effort
was made to specify the hardness by a standard hardness test.

(b) Dimensions of base and insulation against base of supporting table. At first, shock
insulation was achieved by placing the device on a telephone book or several magazines.
Later, rubber feet were found satisfactory, to eliminate the influence of the supporting
table.

(c) Material, dimensions, and weight of cup.

(d) Vertical height of drop of point on cup which comes in contact with base.

(e) Dimensions of groove. (The A.S.T.M. [American Society for Testing Materials]
curved grooving tool, which was developed several years after the original grooving tool
was introduced, is not satisfactory because it does not control the height of the groove
with the necessary accuracy.)
Within a few years after the publication appeared (Casagrande, 1932), the proposed
device was tacitly accepted in many countries as standard. Occasional checks between
laboratories showed that it was possible to duplicate results with reasonable accuracy. But
whenever other materials were substituted for hard rubber, the results could not be checked.
Also, hard rubber blocks of such thickness are not always readily obtainable. A particularly
bad situation developed during the 1939945 war. when even the U.S. Engineer Corps were
unable to obtain hard rubber, so that the Aviation Engineer Battalions were furnished with
liquid limit devices with wooden bases. The development of modern plastics gradually
replaced hard rubber, so that it became increasingly difficult to get hard rubber blocks of
such thickness. Therefore, starting in 1945, the Author reviewed the physical properties of
numerous types of plastics in an attempt to find a material with similar elastic properties
that is manufactured to more rigid standards, and which would maintain its elastic properties
over longer periods than hard rubber. Finally, the Author settled on a fabric-based plastic,
which is manufactured by Westinghouse under the trade name Micarta No. 221, and which
is a type also used for gears. It has a hardness of 113 by the Rockwell “ M ” test. Closely
similar types of plastic are manufactured under various trade names in the U.S.A. and other
countries. It was observed that such bases also had the advantage that they showed much
less wear in the impact area of the cup, so that the frequency of refinishing the surface of the
base could be reduced substantially.
In 1946 comparative tests were made with six liquid limit devices, three with hard rubber
bases (of different vintage) and three with Micarta No. 221 bases, and using a silty clay.
The results of five of these devices varied within a range of only io/0 in water content. How-
ever, one device gave consistently much lower results. Finally, it was found that its cam
was so cut that the cup was being raised to the very edge, instead of holding the cup at
constant height for the last approximately $ in. of travel between the cam surface and the
cam follower. The height cannot easily be adjusted for the very edge of the cam. Therefore,
the usual method of adjustment a short distance from the edge of the cam results in an actual
height of drop greater than 1 cm, and that, in turn, yields a liquid limit that is too low.
In 1949, investigations on the plasticity characteristics of Bearpaw shale were carried
out in the Prairie Farm Rehabilitation Act Laboratory in Saskatoon, Canada, by Ripley
(1950). As an incidental result, two machines with hard rubber bases (from different
86 ARTHUR CASAGRANDE

manufacturers) were compared with two machines with Micarta No. 221 bases. The average
result of three series of tests each were as follows :-

Type of base Liquid limit

Hard rubber 116


Hard rubber 98
Micarta No. 221 108
Micarta No. 221 105

The variations between the two hard rubber bases are obviously excessive, even though
by coincidence their average result compares reasonably well with the results obtained with
the Micarta bases. However, the results obtained with the hard rubber bases may be affected
by other errors in the apparatus, particularly in the height of drop as caused by an unsatis-
factory shape of the cam, as mentioned before and as further explained below.
In order to counteract increasing laxity by manufacturers of liquid limit devices in adhering
to the standardized dimensions, the Author had detailed drawings prepared in 1949, and
which are reproduced in Figs 1 and 2. These drawings were distributed to many soil
mechanics laboratories and manufacturers of soil testing equipment. Without deviating
from the original standards, the Author included in these drawings certain design details
which experience has shown to be desirable, as referred to in the following paragraphs.
The pin is made integral with the cam follower, thus greatly simplifying the construction
of the cam follower as compared to the original design. The pin is resting in grooves in the
movable supporting plate, detail D in Fig. 1.
Some manufacturers have simplified this detail by arranging the pin in such a manner
that the cup cannot readily be removed from the support. The result is that the mixing of
the soil and the cutting of the groove are carried out with the cup attached to the support.
This is not only awkward, but has the following serious drawbacks : (a) The operator cannot
judge readily whether he is holding the grooving tool normal to the surface of the cup. The
Author has watched many technicians who, with the cup on the base, pulled the tool through
the soil holding it at some arbitrary angle without noticing the wrong position of the grooving
tool. This resulted in a serious reduction in the height of the groove. By holding the cup in
one’s hand, with the cam follower upright, at breast height or even higher, one can swing the
grooving tool with one easy stroke downward in such a manner that its position remains
normal to the cup. Some technicians find it easier to hold the tool still and to rotate the
cup upward. One can also achieve good results by rotating the cup upward and at the same
time rotating the grooving tool downward. But the Author has found that a technician would
rather leave the cup attached to the support than go to the trouble of pulling out the pin that
holds the cup to the support. This is particularly true when the pin is screwed in and not
intended to be removed. Only if the cup can be lifted readily from the support and placed
back without removing the pin, can one ensure that the cup will be held by hand while cutting
the groove. (b) Mixing the soil and cutting the groove with the cup attached to the support
will cause dried clay along the edge of the cup to fall on to the base and cushion the impact
of the cup. By removing the cup from the base every time the sample is mixed in the cup
and the groove is cut, one ensures not only better mixing and cutting of a more accurate
groove, but cleanliness of the base. Just before replacing the cup one looks at the surface of
the base and one can wipe off with one stroke of the back of the hand any dry soil that
has dropped on to the base.
In this simplified design, the cam (detail E, Fig. 2) consists of two eccentric circles
instead of the spiral in the original design, which was difficult to manufacture. This design
NOTES ON DESIGN OF LIQUID LIMIT DEVICE 87

?
=
ELEVATION SECTION A-A

GROOVING TOOL -

CALIB.QAllON OF HEIGHT OF FALL J”L” 1949 SHEET 0 OF 2

Fig. 1.
88 ARTHUR CASAGRANDE

f?od=12.7=~2w
Rod=19.0=%”
4

DETAIL E
(Stool)

I : 2 =o
‘w

r* y!-

---f-B
=6*k

DETAIL D IO-24
(Braas)
DETAIL F
(Stool)

DETAIL G

DETAIL H DETAIL J DETAJL K


(Brarr) (Braso) (Brass)

I 0 I 2
i...I.*,l I I
INCHES
SOIL MECHANICS LABORATORY
HARVARD UNIVERSITY
NOTE DIMENSIONS:
ON
First dlmenslon given is in millimeters. LIQUID LIMIT DEVICE-DETAILS
nmeons some roleronco is permitted.
=meono dimension must be precise. JULY 1949 SHEET 2 OF 2

Fig. 2.
NOTES ON DESIGN OF LIQUID LIMIT DEVICE 89

ensures that for the last portion of travel between cam follower and cam the height of the
cup is maintained constant.
To improve the accuracy in adjusting the height of drop, the fine adjustment screw (detail
J in section A-A, Fig. 1; also Fig. 2) was added.

DEVIATIONS FROM THE ORIGINAL STANDARD

The Author has found the most frequent and serious deviations from the standardized
dimensions in the grooving tool. For example, only recently he came across a new grooving
tool (bought from a well-known firm) with a bottom width of 2.8 mm instead of 2.0 mm. But
more often such deviations are due to long wear of the tool, which will cause an increase in
the 2.0-mm width and a decrease of the height of the cut.
On several commercial devices which the Author recently inspected, he found the cam
follower machined in such a manner that the cup rises and then drops in height somewhat
before the cam follower slips off the edge of the cam. Hence, with the usual rate of rotation,
the actual height of drop is larger than that corresponding to the edge of the cam, and the
exact height is uncertain. Furthermore, there is also an uncertainty regarding the calibration
of the height of the drop. For such faulty cams it is usually calibrated for the highest point,
which is too high, but sometimes the calibration is made as close to the edge as possible, in
which case the height of drop is too low. The original design and the simplified design with
eccentric circles (detail E, Fig. 2) assure that the height of drop corresponds to the edge
position. As previously mentioned, it is essential that the height of cup be maintained
constant for a short section at the end of the turn.
Some manufacturers are using cups of thicknesses that are quite different from the original
standard. Although the test results will not be affected excessively for minor variations in
the weight of the cup, variations in thickness of +50%, and the use of different metals results
in objectionable errors in the test results.

THE BRITISH STANDARD FOR THE BASE

During a visit to London in 1956, the Author noticed a liquid limit device with what
appeared to him to be an unusually soft base. When Dr Bishop explained that it was the
British Standard, the Author realized that there might lie the explanation of two mysteries
that had troubled him. During a preceding visit to London, Professor Skempton had told
the Author of measurements in his Imperial College laboratory of the shear strength of
London Clay at the liquid limit of approximately 10 g/sq. cm. That greatly surprised the
Author, because over a period of 20 years his Harvard laboratory had made numerous
measurements of the shear strength of various clays at the liquid limit, using precision direct
shear tests, as well as other methods, and it had been found to range between 20 and
30 g/sq. cm. The other unexplained conflict was the numerical values that were suggested
by Professor Skempton for the boundaries between inactive, normal, and active clays
(Skempton, 1953), which differed substantially from the activities computed by the Author
from test results for various clays. In view of the uncertainty of the determination of the
clay content, which depends greatly on the chemical admixtures used for deflocculating the
suspension, the Author had assumed that the differences were chiefly due to differences in the
technique of the wet-mechanical analysis. But substantial differences in plasticity indices
resulting from different hardness of bases seems now a more logical explanation.
The Author has since learned from Mr D. J. Maclean of the Road Research Laboratory
that when the test was under consideration by a committee of the British Standards Institu-
tion, a device was sent to the U.S. Bureau of Public Roads where it was found to give identical
results with a U.S. machine. The hardness of the base of the British machine was measured
and recorded as a requirement of the British Standard, which thus became, and still is, more
90 ARTHUR CASAGRANDE

stringent than the corresponding A.S.T.M. specification. However, it is clear that liquid
limits determined on both British and American machines are liable to be in error as a result
of variability in the rubber bases, and it is highly desirable from the point of view of interna-
tional standardization that some co-operative efforts should be made on both sides of the
Atlantic to revise and tighten the present specifications.

MEASUREMENT OF THE PERTINENT PROPERTIES OF THE BXSE

There is a serious question whether hardness tests reflect those dynamic-elastic properties
that govern the results of liquid limit tests. The Author believes that a dynamic modulus of
elasticity and its variations with humidity and temperature within pertinent ranges would
be a much better measure for establishing a standard for the base. In the absence of the
results of a detailed investigation of this question, he suggests that both a dynamic modulus
of elasticity and a hardness number be used, and in addition parallel liquid limit tests on
identical clay specimens which are performed at the same time by the same operator and
using machines with identical dimensions.

SHEAR STRENGTH OF CLAY AT THE LIQUID LIMIT

The recent investigations by Norman (1958) include interesting results on the shear
strength at the liquid limit for the five materials tested. The average shear strength at the
liquid limit determined with the Micarta base was 18 g/sq. cm and for the British Standard
hardness base only 13 g/sq. cm. Exceptionally low values of 11 and 8 g/sq. cm, respectively,
were obtained for the Oxford Clay, and when omitting these values, the averages for the
other four clays are 20 and 14 g/sq. cm, respectively. That average of 20 g/sq. cm for the
Micarta base is equal to the lower limit of the range of shear strengths of 20 to 30 g/sq. cm
which have been measured in the past in several independent investigations. Tests by the
Author were all carried out on silty clays which may be comparable to the Weald Clay and
the Brick-earth in Norman’s investigation and for which Norman obtained 21 and 23 g/sq. cm
shear strength with the Micarta base. Thus, using comparable clays, Norman’s results are
approximately the same or only somewhat smaller than those determined at Harvard.
When comparing the liquid limit and the shear strength values reported by Norman,
there appears to be an indication that the clays with the high liquid limits have lower shear
strengths. A similarly ill-defined relationship was observed in some of the Harvard tests.
It is possible that the clays having the lower liquid limits possess a slight degree of dilatancy,
which would explain their higher shear strengths. In the extreme case of a non-plastic or
slightly plastic rock flour the shear strength in a static shear test performed at the liquid
limit is much greater than that for clays.
Other factors that influence the relationship between liquid limit and shear strength are
the thoroughness of mixing and thixotropy. Particularly fat clays are difficult to mix when
starting from the natural water content ; and to obtain a smooth paste may require repeated
mixing on successive days, with rest periods to allow the moisture to distribute itself uniformly.
The influence of thixotropy depends on short rest periods during operation of the tests and
which vary for different operators. Therefore, for comparative studies of the types of bases,
a clay should be used that shows a minimum of thixotropic hardening, and the clay paste
should be prepared from a dried and thoroughly mixed clay powder.

CONCLUSIONS

The liquid limit test as defined by Atterberg, and as performed mechanically with the
liquid limit device, is in reality a dynamic shear test. This is a serious disadvantage because
it does not provide a uniform basis of comparison for fine-grained soils which differ in their
NOTES ON DESIGN OF LIQUID LIMIT DEVICE 91
reaction when subjected to a shaking (dilatancy) test. A simple direct shear test, or an
indirect shear test, e.g., a static penetration test or using the principle of Jiirgenson’s squeeze
test, would eliminate many of the difficulties one faces in the use of the liquid limit device.
Unfortunately, so far none of these tests has been simplified to an extent that it could
compete in simplicity and cost with the present form of the liquid limit test.

REFERENCES
ATTERBERG, A., 1911. “ uber die physikalische Bodenuntersuchung und iiber die Plastizit&t der Tone ”
(” On the physical testing of soil and the plasticity of clays “). Zlzt. Mitt. Bodenkunde, 1 : 5.
CASAGRANDE,A., 1932. “ Research on the Atterberg limits of soils.” Publ. Rds, 13 : 121-130, 136.
NORMAN, L. E. J., 1958. “ A comparison of values of liquid limit determined with apparatus having
bases of different hardness.” Gdotechnique, 8 : 2 : 79.
RIPLEY, C. F., 1950. “ Procedures and equipment used by the P.F.R.A. in the investigation, design and
construction of water development projects.” Tech. Memo. Nat. Res. Council, Canada, No. 17 : 54.
SKEMPTON, A. XV. 1953. “ The ‘ colloidal activity ’ of clays.” Proc. 3rd Znt. Conf. Soil Mech. (Zurich),
1:57-61. ’
TERZAGHI, K., 1926. “ Simplified soil tests for subgrades and their physical significance.” Publ. Rds,
7 : 153-162, 170.

You might also like