Till Fabric Analysis

Source: Andrews, J.T. 1971. Techniques of Till Fabric Analysis. Technical Bulleting No. 6, British Geomorphological Research Group, 43 pp.

Till fabric analysis is the study of the orientation and dip of particles within a till matrix. The definition is restricted by the term 'till fabric' to include only deposits of glacial origin. However, the methods described here are applicable to a study of the fabric of any deposit, regardless of origin. Early geologists noted the tendency for stones in a till to have a preferred orientation related to ice movement; either parallel to ice movement or a transverse particle orientation in certain circumstances. There has been an increasing awareness of the desirability of using descriptive statistics to derive such parameters as mean orientation, the amount of variability, and whether or not the observed till fabric distribution is statistically meaningful.

It is still far from clear what processes or process control fabric orientation and dip. Because of the lack of descriptive statistics, we are as yet unable to ascertain if fabric strength varies between tills of different composition or features of different depositional history. Such questions are of vital importance if till fabric techniques are to provide information, not only on the direction of orientation, but on the process, or processes, that produced specific deposits. The methods outlined here should result, eventually, in a body of comparable information. Many of the techniques in till fabric analysis are similar to those encountered in structural geology, sedimentology, and sedimentary petrology.


Uses and geomorphological implications

The result of any particular study depend to a large degree on the research questions that are being asked and it is only good practice that many of these questions should be formulated before field work begins. The purpose of the study will largely control the details of sampling, methods of statistical analysis, presentation of data, and the conclusions that can be drawn from the study. It is not possible or wise to recommend one particular experimental design. However, it is worth stressing that this freedom of action reduces the potential amount of strictly comparative data.

A review of the literature on till fabrics indicates the following broad fields:

  1. regional analyses of the orientation of till stones, based on a one-site/one-sample basis and very often concerned only with the orientation of the particles,
  2. detailed analyses of a limited number of sections in order to define the sampling problem connected with till fabrics and to observe the variability of till fabric in a single exposure or adjacent exposures, and
  3. studies aimed at the genesis of till fabrics through detailed measurements. This could include an examination of the geometry of the till stones, mechanical properties of the matrix and theoretical considerations.
  4. examination of the sedimentological history of various moraine types or drumlins through a study of the dip and orientation of enclosed pebbles

In many cases it is difficult to categorize studies into one of these groups, and frequently a study concerned with points (1) or (2) will make a contribution to our understanding of till fabric genesis. Measurements taken in such studies can range from the measurement of a axis orientation to the measurement of orientation and dip of all three particle axes and to the measurement of size, shape and roundness. Analysis of data can vary from the use of simple rose-diagrams to more elaborate cartographical devices and three-dimensional vector analysis. Clearly, the procedures of sampling and analysis will vary in direct proportion to the complexity of the stated purpose.

If we could allow for the effect of size, shape and roundness, in short, if we knew how till fabrics are formed, then purpose (1) is satisfactory. However, research suggests that purposes (2), (3) and (4) should be given most attention. At least three methods for till stone orientation have been proposed: these are the 'plastering on' hypothesis; the letting down of material from shear planes; and, thirdly, the orientation of particles by the flowage of the till matrix under hydrostatic pressure. All methods require differential movement either between the ice and the till or between the larger till particles and the finer matrix. Consequently, the character of ice and till is important in considering the response to such flows. Equally important is the effect of changing geometry of till particles even under conditions of unidirectional flow.

The uses of till fabric analysis described above are research oriented and may be separated from the narrower, statistical purpose of a till fabric sample, which is concerned with using the sample as a means of generalizing about the population of till particles at that exposure. In this connection the aim must be to take the smallest sample commensurate with the stated precision of the survey and in the most efficient manner. Field procedures of sampling till particles are, unfortunately, not very rapid.

Experimental design

A single sample (of say, 50 stones) at an exposure does not provide sufficient data for the researcher to generalize about the population of till stone orientations and dips at the exposure. It is analogous to selecting one city block and from the answers to a questionnaire making pronouncements on the voting habits of the entire city. The single sample approach provides information on the sample and little else. Our single sample cannot provide a picture of the entire section - for example, the site may be adjacent to a large, hidden boulder or changes in the build-up of the till sheet may be reflected by changes in orientation of till stones between the base and top of the exposure. It is, therefore, extremely important to utilize design of experiment concepts. Information on these concepts are included in most statistical textbooks, and extended discussion is not needed here. There is very little evidence of experimental design in most papers on till fabric.

In order to understand the purpose of a sampling scheme it is necessary to differentiate the different types of 'populations'. Normally, till fabrics are taken from the vertical face of an exposure and the target population can be defined as the orientation and dip of all elongate stones in a particular deposit. This population differs from the hypothetical population which consists of those elements of the population that have been removed and are not available for sampling. Conversely, the existent population consists of till stone orientations and dips that occur in the exposure. A number of factors may combine to limit sampling to the available population. For example, the difficulty of sampling a 30 m vertical exposure could well result in a sampling design that was limited to the lower and upper 3 m of the till. However, such a design is biased and should not be used to predict the behavior of the existent population. The investigator might wish to make such an extrapolation on the basis of his own judgement but this must be clearly stated. It can be argued that many existent populations cannot be adequately sampled because of the difficulty, time or expense, how-ever, the following statement should be considered:

Two other important considerations include randomization and replication of samples. Randomization of the' experiment is required if probability statements about the entire population are the object of the study. If the sampling plan is so arranged by the investigator to include 'typical elements' then the design is termed a 'fixed effect' design. Judgements about the till fabric population from such a scheme can be made by the investigator on the basis of his geomorphological knowledge, they cannot be made into probability statements. The variance of till fabrics is such that several observations should be made at each individual sample site. The majority of till fabric studies consist of samples of size 50 to 100 determinations. Designs of experiment allow for such replication.

In the above discussion it has been assumed that the sample of orientations or dips is part of a normal population. Special problems occur because a measurement of orientation is not unique, that is 010 and 190 degrees can only be distinguished if the sense of stone imbrication is included. Rrather than take a sample of 100 till stones at one site, another approach would be to take 10 samples of 10 pebbles, calculate the means (in effect one method of normalizing samples) and use these 10 means as our replication. This method reduces the variance and thus increases the power of the design.

Field Procedures

Field procedures cannot be isolated from the purpose of the study and there is no method that can be advocated above another. Three levels of inquiry can be set-up. The first and most common approach includes the measurement of all necessary parameters at the exposure. The second approach is a laboratory study of macro-fabrics taken from orientated till blocks or cores, or by the re-orientation of marked pebbles. Thirdly, blocks or cores of till can be impregnated with resins or epoxies, allowed to set and then thin-sectioned and measurements performed under a microscope.

Clearing the exposure

Regardless of the method to be used, it is necessary to clean and excavate the exposure. Most sampling has been undertaken on vertical exposures; the outer 0.3 m should be removed in case of disturbance. If no cuts are available a trench must be dug. Soil creep, solifluction, frost-heaving and farming could have disturbed the topmost portion of the exposure. Till fabrics from the upper 30 to 50 cm may show some degree of disturbance. Although orientation is not affected by freeze-thaw, dip is usually altered.

Sampling the site

Till stones have been sampled on both vertical and horizontal faces, but the problem of which method introduces the least bias has not been adequately considered. Intuitively, it would seem that sampling a horizontal exposure would be the best method, especially if the till stones had dips within +/- 20 of the horizontal; there is a great temptation when sampling a vertical face to select pebbles that are projecting out of face. Ideally, both surfaces should be sampled and an analysis of variance test run to establish if there is a difference in the methods. If the vertical face is gullied, or has marked buttresses, samples can be taken on the various faces to test for sampling bias.

If measurements are to be taken in the field, the cleared face should be slowly excavated with a knife or trowel until a suitable stone is located. If there is sufficient silt/clay in the matrix the stone can be carefully removed leaving its cast in the till. The pebble is then examined and reinserted for measurement. Sampling in a coarse friable gravel, or gravelly-till, is extremely difficult as the removal of the pebble frequently results in a collapse of that portion of the face. The marking of the stone and re-orientation in the laboratory is advocated. Conversely, the matrix can be impregnated with a setting resin thus enabling the entire block to be removed and examined in thin-section.

It is common practice to avoid pebbles that lie adjacent to large boulders, or that touch each other or that have particularly high dips. Such procedures are the product of value-judgements on the relevance of such situations and automatically bias the data. For example, the restriction of dips to + X0 from the horizontal means that any three-dimensional test on the randomness of the fabric will, in all probability, be rejected.

Definitions and measurements of axes and other parameters

Because stone size and geometry can influence the location of a pebble in three-dimensional space, stone size and shape should be measured together with the appropriate angular measurements. In till fabric studies all three axes of a pebble should, ideally, be measured.

Any study of the orientation and dip of a, b, or c axes must be internally standardized. If effective and meaningful comparisons are to be carried out on till fabrics from different features, studied by two or more people, it is also highly desirable to employ standard procedures. Sometimes the a axis is defined as the longest axis but such a definition ignores the interaxial relations. In terms of the dynamics of fabric genesis, a method that takes into account the maximum and minimum projection planes of the till stone is preferred. Place the pebble on a flat surface and allow it to attain a stable position, normally it will rest with the maximum projection exposed. The b axis is defined as the shortest axis across the maximum projection plane with the a axis normal to it. If the pebble is turned until the minimum projection is revealed then this defines the c axis. A common problem is the correct angular definition of the a axis of a rhombohedral shaped pebble.  If the term 'longest' is accepted then the axis lies along a1 but such a definition contravenes the definition adopted above which is a2. Note that b2 is the shortest diameter in the maximum projection plane not b1. The length of the axes can be measured with a tape or preferably, sliding calipers, certainly to within +/- 2 mm.

If the three pebble axes are measured then it is appropriate to use the Zingg classification which delimits four broad shape groupings: oblate, spherical, bladed and rod-shaped.

  • definition of stone roundness
  • The roundness of particles in a moving media is of potential significance as it could effect the degree of drag around the particle which might be reflected in the orientation and dip. Roundness refers to sharpness of the edges and corners of the clasts and has no connotation with shape. Visual estimate charts of roundness are satisfactory. Frequent use is made of the Cailleux index where r is the minimum radius of curvature of a corner in the maximum projection plane, the length of the a axis is L and the index = 2r/L(l000). The index varies between 0 and 1000. Measurements are performed in the field by comparing pebble roundness to a target consisting of semi-circles of varying radii.

    A difficult decision in till fabric analysis is what limit to impose on the a/b ratio of a pebble before it is considered suitable for bearing and plunge to be measured. In other words, how long has the long axis to be in relation to the intermediate axis? A common limit is 2 :1 but ratios of 1.2 : 1 must be accepted at some sites. There is no simple answer to the question as a great deal depends on the type of bedrock fragments in the till. For example, slates tend to form excellent elongate rods but they could also fracture into nearly square plates. Granite pebbles are often nearly equi-axial and consequently difficult to measure. In an ideal situation measurements might be restricted to a particular rock type (= shape).

    When selecting a particular a : b ratio, bias has been introduced and our data refer only to pebbles with an a : b ratio above the specified threshold. If the maximum projection plane is used, this problem is not so severe that good results cannot be obtained from particles with axes a = b. Such material should be included by measuring bearing and plunge of the c axis.

    Most authors place upper and lower limits on the size of particles that are measured in macro-fabric studies. The effective lower limit is imposed by the crudeness of field techniques when material is less than 8 mm. Smaller particles should be examined in thin-section after suitable pre-treatment. The upper limit appears to be about 100 to 120 mm. Studies have, however, been made on particles of 0.5 m and over. Size seems to have some constraints on orientation. Ideally, fabrics from the entire size spectrum of the deposit should be examined if the aim of the study is related to the genesis of till fabrics.

    Measurement of orientation and plunge of pebble axes

    The discussion of angular measurements will concentrate on the direct field measurement of the orientation and plunge of pebbles through measurements on the a axis, and secondly, on the measurement of particles in the laboratory after they have been collected in the field. Of the two methods the direct field measurement is most common.

    The degree of sophistication and number of parameters measured depends on the object of the research survey. Often, all that is required by the investigator is information on the orientation of pebble a axes. Such a research objective is less demanding of time than a survey which might, for example, be examining the intra-exposure variability of till fabric in which dips are also measured.

    Once the face of the exposure has been prepared, stones are gently removed from the till matrix and examined to ascertain the position of the a axis. A non-magnetic object, such as a knitting needle or brass rod is then inserted in the cast left by the stone and oriented parallel to the a axis. The bearing can then be measured by a compass or orientometer. The selection of the compass should be done with care. A liquid-filled one (e.g. Suunto fabric compass) gives quicker readings than a pivot-bearing model. Orientations should be measured as precisely as possible to minimize the progressive loss of information that comes from processing the data statistically or more critically, on diagrams. The fact that readings are attempted to +/- 1 does not mean that this is the level of confidence attached to defining the a axis. The appropriate corrections should be applied for magnetic declination. The orientometer measures bearings relative to the trend of the exposure. A protractor mounted on a ruler is set normal to the face and the orientation of the rod is then read as x0 left or right. These can then be converted into true bearings in the office.

    The measurement of dip is more difficult and has higher errors attached to it. The non-magnetic rod should be replaced in the cast to exactly duplicate the plunge of the a axis. The plunge is then read by an inclinometer. Several makes of compass have small inclinometers that can be read to +/- 5. One of the main problems relates to that of keeping the rod sufficiently firm so that a reading can be taken. Particularly in tills of high sand/ gravel content such a task is difficult. The measurement of dip is easier if pebbles are taken from a horizontal face (as opposed to vertical) as then the problem of pebbles falling outis overcome.

    The problems of replacing a pebble, or its analog into its cast, of correctly determining the position of the a axis, and the desirability of examining the data three-dimensionally, indicate the probable advantages of taking the stones to the laboratory. If this is done the location of the maximum and minimum projection planes and their sizes can be recorded.

    A template (Figure 4, inset) is held parallel to the face of the exposure and leveled by means of a small spirit level. The exposed surface of a pebble is then marked along the + of the template (Figure 4) and a dot is placed in the lower, left hand corner of the +. The stone is then removed and can then be numbered or coded by waterproof ink. In the laboratory the stones are re-orientated into their former positions. The stone is mounted on modelling clay at the center of a circle. A transparent plastic plate that has been etched with cross-hairs (Figure 4) is aligned parallel to the bearing of the outcrop. The stone with its + marking, is viewed through the plate, and is re-aligned to its previous position. The major and minor projections of the pebble are best determined prior to setting the stone on the modelling clay and can be marked. The orientation of the a and c axes are determined by laying a straight edge across the stone and noting the intersect on the circle. Dips are measured by a protractor and plumb bob attached to the straight edge.

    Surface boulders have been used to determine the direction of glacial movement. As the boulders are usually embedded in till, measurements are performed on the apparent a axis by using a liquid-filled compass (thus analogous to measurements of orientation in thin-section).

    Micro-fabric studies of till are less numerous than on the macro-element of the matrix. Orientated blocks of till of size 6 x 6 x 2 cm are carefully cut from the side of a trench in the deposit. The orientation, say north, is marked on the sample and a horizontal line is added. In the laboratory, the sample is impregnated with a commercial casting resin. Problems of complete impregnation can be eliminated by placing the sample under partial vacuum. The sample is allowed to dry and thin-sections are then cut. Generally orientations are measured from the horizontal cut, but ideally all three planes should be examined.

    Booking and displaying results

    It is preferable to book the results on standard forms that are designed with computer processing of the data in mind. Two basic forms are involved - the first includes details of the site whereas the second notes particulars of individual stones at each sample site.   They should, of course, be altered to accord with individual particular research aims. It is extremely important that all pertinent information is recorded at the time of the investigation, especially if the area is remote and cannot be revisited. A viable data form(s) is important to the success of the project and conditions the usefulness of the data for later retrieval. For example, the till fabric analysis form includes space for important comments on the method used (including the equipment), whether the face was vertical or horizontal, what the units of measurements (cm, mm, etc) plus date, collector, and any restrictions imposed (e.g. a : b ratio of 3:1).

    The graphical display of the distribution of orientations and/or dips is an important way of communicating the information. A number of basic diagrams are in use. The selection of an appropriate graphical model will be influenced by the objectives of the study, the precision of the data, the number of pebbles per sample, and the number of samples. It should be borne in mind that it is difficult for readers to visually compare and contrast a large number of fabric diagrams. The perception. of changes in fabrics is difficult when the number exceeds about 15.

    The most common form of the rose-diagram portrays the number, or proportion of pebbles in different azimuthal classes.  It is easy to construct and conveys a strong visual message, but it results in a loss of information and the visual impact can be altered by changing class intervals and the starting point.   For example, instead of grouping orientations into 20 class intervals starting at 040, a 10 interval might be used starting at 035 and possibly would result in a quite different pattern.  Also the repetition of the frequencies through 180 tends to magnify the strength of the fabric.   To provide a partial counterbalance to this last point it is more meaningful to plot the data with the sense of direction given.  Dips can be plotted in a similar way.