DIVISION S-6—SOIL AND WATER MANAGEMENTAND CONSERVATION

Soil Anisotropy and Its Relation to Aggregate Stability1

B. CAGAUAN AND G. UEHARA2

ABSTRACTFor kaolinitic soils containing more than 5% free iron oxide,

variations in percentage aggregate stability can be related tosoil anisotropy. Soil anisotropy, as used in this paper, refers tothe degree of particle orientation. In thin sections, increasingparallel orientation of clay particles within soil aggregatesallows increasing reaction of the soil material to polarizedlight. For soils having essentially identical mineralogies, theamount of light transmitted through crossed polarizers can berelated to particle orientation or soil anisotropy. The amountof transmitted light, which is used to measure soil anisotropy,can be measured with a photometer. When percentage aggregatestability was plotted as a function of photometer readings, asignificant curvilinear relation was found to exist between thesetwo variables.

IN AN EARLIER PAPER Uehara et al. (9) described a seriesof soils of essentially identical mineralogies but which

differed greatly in clay fabric. At that time, it was postu-lated that structural stability of these systems might beassociated with fabric type. Interpretation of soil fabrichas recently been greatly facilitated by the published workof Brewer (2, 3). An important part of soil micromor-phology concerns itself with the concentration and arrange-ment of clay particles in a soil system. Soil engineers havelong recognized the significance of particle arrangement tothe engineering properties of soil materials. Mitchell (8)has described a nearly identical method for measuring clayfabric and has related particle orientation to a number ofengineering properties. Lambe (6, 7) has given a thoroughtreatment of the relationship between clay structure and itsbehavior. In glacial lake clays, Wu (10) noted that micro-

structure of clays varied with the mode of deposition andmineralogical content, and that strength characteristics weredependent upon the microstructure of the clays.

Soils may be anisotropic in many ways and to manythings. In this paper, the use of the term "soil anisotropy"will be restricted to the optical anisotropy of the natural,undisturbed soil pedological unit. Optical soil anisotropymay vary with the mineral composition of a soil. However,for a pedological system composed mainly of clay-size par-ticles (in this case kaolinite and noncrystalline free ironoxide in unvarying amounts), optical anisotropy will varymainly with particle orientation or the intraped organiza-tion of the soil constituents.

Since most soils are composed of anisotropic, crystallinecomponents, no soil is completely isotropic. The minutesize of the crystallites, however, makes it impossible forthem to react to polarized light. For this reason, when thecrystallites are arranged in a completely random fashion,the system is nearly isotropic. Such a system is usually de-void of any pedological organization and might be consid-ered the starting point for the development of soil aggre-gates.

Maximum soil anisotropy is attained when all crystallitesin a soil unit are oriented in one crystallographic direction.This condition imparts to the mass the characteristics of asingle crystal.

The two situations just described constitute end membersof a continuous sequence in soil optical anisotropy. In realsystems, these end members never exist except in dimen-sions on the order of a few microns. Clay skins or illuvia-tion cutans are prime examples of the latter case.

In brief then, other factors being equal, the reaction ofpolarized light with the soil mass is a function of thedegree of particle orientation. Qualitatively, one mightrefer to a system with random particle orientation (edge toface) as being weakly anisotropic and to another with pre-ferred orientation (face to face) as strongly anisotropic.This comparison is possible only when the sample is viewedand compared under crossed polarizers. Quantitatively, therelative amount of light transmitted through crossed polar-izers becomes a measure of the degree of soil anisotropyand herein lies the definition of this term.

CAGAUAN AND UEHARA: SOIL ANISOTROPY AND ITS RELATION TO AGGREGATE STABILITY 199

Fig. 1—Three soil-fabric types encountered in this study. X125

MATERIALS AND METHOD

Percentage aggregate stability as a measurable quantity can berelated to such factors as free oxide, clay type and content, organiccarbon, silica, carbonate, exchangeable sodium, etc. The essence ofthis paper is to demonstrate that soil anisotropy as denned earliermay be included in this list. This is a relatively simple matter ifone selects samples that vary only in soil anisotropy. To approachthis ideal situation, samples were collected from soils of the low-humic Latosols from the Island of Oahu. The soils of this GreatSoil Group which have been described in the Soil Survey of theTerritory of Hawaii (4) are fairly uniform.

The amount of light transmitted through crossed polarizersdepends on the amount and kind of minerals, thickness of theslide, etc., as well as particle orientation. Thorough mineralogicalanalyses by X-ray and differential thermal method were carriedout to establish similarity of mineralogies in these samples.. X-raydiffraction was used primarily for qualitative and semiquantitativeanalyses of soil minerals. This established the predominantly kaolinnature of the sample. Quantitative comparison of the kaolin con-tent was made by differential thermal methods. Free iron oxidewas determined by the method described by Aguilera and Jackson(1); and organic carbon, by chromic acid digestion.

A thorough micromorphological examination showed the pres-ence of three types of soil fabric as shown in Fig. 1. Type a wasthe only one considered in this study. In type b, the concentrationof anisotropic material on ped faces and channel walls made itimpossible to obtain meaningful photometer readings. In type c,high concentrations of birefringent gibbsite disallowed use of pho-tometer readings as a measure of particle orientation. Thus, onlythose samples with similar mineralogies displaying varying degreesof particle orientation, uniformly distributed through the soil mass,were utilized. This feature is shown in Fig. 2. One can readilysee that for a given photographic exposure time more light istransmitted in the samples possessing higher particle orientation.In each instance, however, the anisotropy is fairly uniformly dis-tributed throughout; rotation of the microscope stage does notalter the photometer readings in these systems.

No index mineral of sufficient size was present in these soilsso that approximation of slide thickness was made by observationof the interference color of the whole mass. These soils were quiteuniform in color and depending on slide thickness, the interferencecolor ranged from red, reddish yellow, to yellow. The reddishyellow color was selected for all comparisons. The slide thicknesswas believed to be near 25^. A comparison of photometer read-ings from slides of varying thicknesses, however, showed a greaterdependence of reading on particle orientation than on thickness.For example, overall readings as a function of particle orientationranged from 5-25 units whereas in readings from a number ofslides of the sample, readings never varied over 2 units from themean.

A Zeiss microscope provided with an exposure meter was usedfor all anisotropy measurements. Each photometer (exposure meter)reading represents an average of 25 readings from at least threeslides.

For aggregate analysis, the 2- to 3-mm fractions were subjectedto wet sieving for 30 min and the percentage of dispersible aggre-gates greater than 0.25 mm retained on a sieve was used for cal-culating percentage aggregate stability. Each aggregate stabilityvalue is an average of 10 determinations.

Fig. 2—Photographs showing increasing transmission of lightthrough cross polarizers as a function of increasing soilanisotropy (particle orientation). X50

RESULTS

Table 1 summarizes the effect of free iron oxide, organicmatter, and soil anisotropy on percentage aggregate stability.These data are included to demonstrate that the first twovariables were not responsible for the observed result inFig. 3. Clay content was considered an important variable,but values reported by Kawano and Holmes (5) andMatsusaka3 give figures near 80% for the same soils. Theirdata suggest that clay content varies only slightly in thesesoils.

A thorough mineralogical study of the whole soil andits respective clay fraction showed that all samples selectedfor this work had essentially identical mineralogies andthus, soil mineralogy was not treated as a variable in thesesoils.

A test for curvilinearity confirmed this trend and Fig. 3graphically depicts the relationship between percentageaggregate stability and soil anisotropy.

DISCUSSIONThe results of this investigation are, in general, contrary

to other findings. It is generally believed that the edge-to-face arrangement constitutes a more stable system than theface-to-face arrangement. The apparent discrepancy isattributable to two major factors: (i) the mineralogicalcomposition of the system and (ii) its history of develop-ment.

3 Matsusaka, Y. 1952. T h e r m a l and ionic characteristics ofHawaiian soils. M.S. thesis, Univ. of Hawaii.

200 SOIL SCIENCE SOCIETY PROCEEDINGS 1965

Table 1—A comparison of the effect of 3 soil variables onaggregate stability

Correlation coefficient Degrees of fFree Fe2 O3Organic cSoil anisotropy

-.490+ .329+.621*

121212

* Significant at the 5% level.

Mineralogically, the samples studied were composed offine grained kaolinite and free iron oxide. The combinedtotal of other minerals such as anatase, gibbsite, quartz,feldspar, and mica never exceeded 10% by weight. Thus,for all practical purposes these samples can be consideredto be a simple kaolinite-free iron oxide mixture or, inpurely physical terms, plates and cementing agents. Quan-titative estimation by differential thermal analysis gavekaolin contents ranging between 60-80%. Free iron oxidecontents of 8-20% were measured.

There is reason to suspect that had the kaolin componentbeen replaced by montmorillonite, the results would havebeen quite different. The same can be said if a purely kao-linite soil with no oxide had been used.

More important than the mineralogical aspect, however,is the one of time and history of formation. It should beemphasized that the results of this study apply primarilyto natural soil aggregates (peds) which have been pedo-logically organized over geologic time. Remolded soils, orsoils that have been intensively cultivated, should receiveseparate study.

As a final inquiry, one might ask how differences in soilanisotropy might arise in soils of nearly identical mineral-ogies. Those soils showing the least degree of particleorientation were collected from relatively dry areas, whereasthe strongly anisotropic ones were collected from areas ofhigher rainfall, but with a pronounced dry season. Itappears that alternating wetting and drying favors particleorientation and high aggregate stability. Soils developedunder still higher but uniformly distributed rainfall pos-sessed isotropic soil peds coated with illuviated clay films(Fig. Ib). The latter type was omitted in this study becauseof difficulty in measuring its soil anisotropy. In general,this type displayed extremely high aggregate stability. Asa rule, soil anisotropy increased with increasing rainfall, upto a point, and also increased with increasing depth withina profile, reaching a maximum in the B horizon.

Soil anisotropy apparently not only affects soil struc-ture, but appears to control other soil properties as well.Although the measured data are few, the moisture contentby weight at a given tension increases with increasing soilanisotropy. It would be both interesting and desirable toascertain the importance of particle orientation or soilfabrics in general to soil physics as a whole.

100

98

96

b'4

3 92t-t/>w 90

o£ 88oe>1 86

Y = 71.26 +2.9253X- .O8O44X'

•t = 4.48**

6 8 10 12 14 16 18 20PHOTOMETER READING (ANISOTROPY)

22 24 26

Fig. 3—Dependence of aggregate stability on soil anisotropy.

ACKNOWLEDGMENT

The authors are indebted to Dr. Paul Day, University of Cali-fornia, Berkeley, for bringing to their attention a number of perti-nent references in the engineering journals and for his helpfulcriticism of the mauscript.