Critique of Archaeological Reason
5. Excerpts and summaries

Michael J. O’Brien and R. Lee Lyman

Laerke Recht – October 2015

Back to top: Michael J. O’Brien and R. Lee Lyman

Excerpts and extended summary of Obrien & Lyman 2002 Seriation Stratigraphy And Index Fossils

Back to top: Michael J. O’Brien and R. Lee Lyman

Extended summary of Obrien & Lyman 2002 Seriation Stratigraphy And Index Fossils

This highly theoretical book examines relative dating methods used in archaeology, specifically those of stratigraphy, seriation and index fossils. The authors offer a critical review and explanation of each method, along with a useful historical background of each. This background is crucial for the understanding of why we proceed the way we do in archaeological practice today, and for understanding where some things might have gone ‘wrong’ or gotten confused. By going through the merits and pitfalls of each method, the authors show that relative dating must remain a core part of archaeology even in modern times of various absolute dating methods.

The focus throughout the book is on archaeological history in America (or among ‘Americanists’, as the authors prefer), but is very relevant for archaeology anywhere.

The dating method chosen by the archaeologist has serious implication for how time is measured - continuously or discontinuously - and hence for archaeological interpretation. The sound measurement of time is one of the most fundamental tasks of archaeology. A continuous measurement of time allows for the detection of gradual change, whereas a discontinuous measurement of time will tend to see change in the biological term ‘punctuated equilibrium’. The conflation of different means of measuring time leads to chronological errors.

The scales used to measure time can be classified as nominal (equal and unranked, not useful for chronology), ordinal (chronologically ordered), interval (units of equal length of time or equal length of time between units), or ratio (with a fixed ‘zero’ point).

The units for measuring time are either ideational (meaning arbitrary, created by the archaeologist) or empirical (already existing, waiting to be discovered). Ideational units are thus not ‘real’, although they measure something real, time, “But the units we use to subdivide time into segments - centuries, years, hours, nanoseconds, and the like - are ideational units. Keeping one’s ideational units separate from empirical units, which the former are meant to measure, is critical to any study of change.” (p.22).

Archaeologists create ‘types’ in order to understand and classify archaeological material. Different types serve to answer different questions - for example, functional types are made to answer questions concerning the function of an object. Chronological or historical/temporal types are the ones useful for measuring time. Types are not ‘real’: they are ideational units created by the archaeologists. The objects placed in the units are real, though.

Seriation: Seriation is divided into three different kinds: phyletic seriation, frequency seriation, and occurrence seriation. The two latter are expanded version of the first. They all measure time continuously.

Seriation creates a linear order of things that are similar to one another. The sequence thus created can measure time, but does not necessarily or inherently do so; this must be tested by other means. Also, seriation does not on its own tell us which ‘end’ of the sequence is older and which is newer; this must be confirmed using supplemental methods, such as stratigraphy.

In seriation, objects are placed in an order based on formal properties or attributes based on similarity. For the seriation to measure time, the assumption is that there is historical continuity - in other words, that the resemblance between the objects reflect proximity in time (rather than, for example, in space). Once historical continuity has been established, heritable continuity must also be so. Heritable continuity implies that one object directly develops from another, and that the similarity is thus explained, while historical continuity only asserts that the objects came one after another, not that they are necessarily ‘related’. As can perhaps be guessed, much of this method has its origins in biology. Characters of objects can thus be seen as either analogous (in biology, this would mean having developed parallel but not actually being genetically related) or homologous (implying a genetic relationship and sequence). In seriation for chronological purposes, we are only interested in homologous traits.

Frequency seriation and occurrence seriation use assemblages of objects rather than single objects. Within these assemblages, they add to usual seriation the attributes of whether something occurs or not (occurrence seriation / quality) or the amount of times something occurs (frequency seriation / quantity).

Stratigraphy or stratigraphical excavation: Stratigraphical excavation has its roots in geological observation. Stratigraphy or stratigraphical excavation is not very clearly defined and leads to confusion. Stratigraphy must be carefully separated from the principle of superposition. Superposition is the deposition of material in ‘strata’ (each representing one ‘event’, not a culture, as is often assumed), with the lowest stratum being the oldest and the top one being the newest. Stratigraphy only tells us the sequence of deposition, it does not give us the age of the contents of each stratum. Stratigraphy is a discontinuous way of measuring time.

Index fossils (cross dating): Like stratigraphy, index fossils have their origin in geology, as the name also suggests. In geology, index fossils are used to correlate layers across space, and similarly so for artifacts in archaeology. Index fossils are objects with “extremely narrow temporal distribution” (p. 187); they are secure chronological markers from one site that can be used to date specific contexts with the same type of object at other sites. As such, this dating method also measures time discontinuously.

Back to top: Michael J. O’Brien and R. Lee Lyman

Excerpts from Obrien & Lyman 2002 Seriation Stratigraphy And Index Fossils

defining archaeology and its goals p. 1      ...we would venture to say that placing archaeological specimens, commonly referred to as artifacts, in their proper chronological position is probably the most fundamental exercise in archaeology. ... All these aspects of artifacts are important components of archaeological inquiry, but at its core archaeology is the study of change.
p. 139      There is no getting around the fact that digging, or as archaeologists prefer to say, "excavating," is central to the discipline.
defining terms related to time and archaeology date and age
pp. 5-6
     Throughout this book we use the term dating to refer to "the placement in time of events relative to one another or to any established scale of temporal measurement" (Dean, 1978:225). It has been suggested that "a date is a specific point in time, whereas an age is an interval of time measured back from the present" and the former term, "when used as a noun, carries a connotation of calendar years and a degree of accuracy that is seldom appropriate" (Colman et al., 1987:315). The latter term refers in part to the "chronological sensitivity" (Dean, 1993:60), or temporal resolution, of the dating technique used. We use the terms date and age in a manner similar to that suggested by Colman et al. (1987), though we do not always find this possible given the chronological sensitivity of the dating methods we discuss.
p. 6
     An event can comprise the manufacture of an arrowhead, the use of a ceramic vessel, the deposition of a bone awl, or the caching of a tool kit (we provide a more detailed consideration of the concept of event in Chapter 2). Knowing the age or date of archaeological events, we can determine to various degrees of resolution the succession of events, the duration of an event, the simultaneity of events, or some combination thereof. We can also, if we so desire, determine the rate, or tempo, of change over particular spans of time and perhaps even detect the kind, or mode, of change.
relative and absolute
p. 8
     What we treat here often are referred to as relative dating methods, meaning that they measure time but with no indication of the amount of time that elapsed between each pair of events and no indication of when on a calendric scale the events occurred. Relative dating methods provide only a chronological sequence of events. In contrast, absolute dating methods, sometimes referred to as chronometric methods, yield the amount of time that elapsed between each pair of events as well as a calendric date indicating when each event occurred and perhaps each event's duration as well.
p. 144
     We need first to examine what we mean by certain terms. The most appropriate place to begin is with the term stratum, which we define as a three-dimensional unit of sediment of any origin that represents a depositional event and is distinguishable from other such units (Lyman and O'Brien, 1999; Stein, 1990). Importantly, this definition emphasizes that a stratum arises from a single depositional event; inclusion of the words "depositional event" denotes that all of the sedimentary particles, including artifacts, comprising a stratum were deposited more or less contemporaneously.
relative and absolute dating p. 9      Relative dating methods produce relative dates, or a sequence of events placed on an ordinal scale. Event A can be said to be older than event B, but we cannot say how much older A is than B; that is, we cannot say when on a calendric scale A and B occurred. Absolute dating methods produce absolute, chronometric, or calendric dates, or sequences of events aligned on a scale of units in which each unit is of the same duration as every other unit. If event A is determined to be 10 years old and event B is determined to be 20 years old, then we can say that event B is 10 years older than event A, and we can also say that B is twice as old as A. Of course, we can only say such things if each of the units, in this case solar years, on the scale are of precisely equivalent duration. It should be clear, then, that it is critically important to keep relative (ordinal scale) time and dating methods distinct from absolute (interval scale) time and dating methods during analysis of the archaeological record. Archaeologists have not always kept the two separate, resulting in such things as the Radiocarbon Revolution (Renfrew, 1973; see also Taylor, 1996), when the absolute dates assigned to various events dated by relative dating methods were found to be far off the mark.
p. 13      The decline in use that relative dating methods, especially seriation, have experienced in Americanist archaeology has in large part been the result of a confidence in methods that seem to be more scientific than the three discussed here. In our opinion much of this confidence is misplaced. Simply because an absolute date is derived by way of chemistry and physics does not make it more scientific than a relative date derived by way of seriation. If the hallmark of scientific research is that its results are testable, then relative dating is readily subsumed under the umbrella of science.
continuous and discontinuous time pp. 9-10      Another distinction that should be kept in mind is of that between viewing time as continuous and viewing it as discontinuous. Both views have played important roles in archaeology, though the distinction between them has rarely been noted. Conflation of the two has caused severe problems that continue to plague archaeological analysis. But, you might think, a clock does not stop; thus time can hardly be discontinuous. Time, you say, is like the flight of an arrow (Gould, 1987): unidirectional and never stopping (ignoring, for the sake of the metaphor, the force of gravity on the arrow's flight). What, then, is meant by the terms continuous time and discontinuous time? In short, we are distinguishing between two different methods and results of measuring time. Time is measured as a sequence of unique phenomena that mark moments in time, whether those phenomena are the shifting ratio of decayed to nondecayed atoms of an isotope, the shrinkage of fins on an automobile or fish, or the altered positions of planets and stars. Time, in other words, is measured relative to change in, or the alteration of, phenomena.
p. 10      Time's flow in the sense in which archaeologists are interested in it and in terms of how they measure it - culture change - is rendered discontinuous. Throughout the book we touch on the problems that result from dating methods that measure time discontinuously.
p. 183      ... if one views change as continuous and gradual, such as Ford did, then superposed materials are supplementary rather than primary to the measurement of the rate of culture change; the role of superposition is to indicate the direction of the flight of time's arrow and to confirm that the types constructed are measuring time.
linear and cyclical time p. 14      First, recalling time's cycle, time can be measured on the basis of cyclical events, such as the movements of the sun and moon; such events form the basis for modem as well as ancient calendars. Second, time can be based on certain constants such as the speed of light (light years) or the vibration rate of quartz crystals. Third, recalling time's arrow, time can be based on successive and/ or cumulative changes in something, for example, in the decay of radioactive substances such as 14C or in changes in artifact form. Fourth, time can be based on the stratigraphic observation of superposed phenomena.
archaeological types p. 24      If types are constructed in particular ways, they successfully measure time as a continuum and can be considered to comprise chronological types. But if they are constructed in other ways, they may measure time cyclically, discontinously, or not at all.
p. 26      How do we construct chronological types? The answer is fairly straightforward: Determine the characteristics or attributes of objects that change over time. ... In other words, chronological types are constructed by selecting attributes that either themselves change through time or the combinations of which change through time.
p. 29      To be of chronological use, types, rendered as particular combinations of attributes, should have continuous distributions in time, and the period of time over which they occur should be fairly short. In other words, each type should have occurred only once, and it should have disappeared after a relatively short life (Fig. 2.2). That is, chronologically useful types cannot reappear at a later date.
p. 51      Although we have not called attention to it, we have maintained a strict distinction throughout the discussion between type as an abstraction - an ideational unit - and the actual objects placed in a type.
seriation p. 60      Marquardt (1978: 258) defined seriation as "a descriptive analytic technique, the purpose of which is to arrange comparable units in a single dimension (that is, along a line) such that the position of each unit reflects its similarity to other units." Seriation is a descriptive method that simply orders things in a row or column.
p. 60      Seriation creates a linear order, but that order only tells us that the odds are good that two adjacent things are more alike than either is to things farther up or down the line. Seriation most definitely is used in archaeology to measure the passage of time, but whether or not the order of units created by a seriation is a sequence in the sense that it actually reflects the passage of time is an inference; it is not axiomatic.
p. 61      That is, seriation is based on intrinsic properties or attributes of the artifacts and not on their relative vertical positions in a column of sediments; the last is an extrinsic property or attribute.
p. 66      The principle of ordering dictates the procedure: place the objects most similar to one another adjacent to one another; as similarity between objects decreases, increase the distance between them. The inference that the order is chronological is founded on the principle of historical continuity. Whether or not the order we produce is in fact a chronological ordering of forms - a temporal sequence - and thus measures time is an inference that must be tested with independent data such as finding the same ordering of forms in a sequence of superposed sedimencs. If the ordering is in fact chronological, and thus denotes historical continuity, then the next problem is to determine if the sequence is also a lineage, a line of heritable continuity. The two - heritable continuity and historical continuity - are not necessarily one and the same.
p. 114      For any seriation technique - phyletic, occurrence, frequency - to work, the rate of change within a lineage must be gradual. Although such cannot be assumed literally, it is precisely such an assumption that allows the positioning of like adjacent to like and unlike to be placed some distance apart in an ordering. But do not be misled by the word "gradual." Characterizing the rate of change as "gradual" does not preclude rapid change or fluctuation in the rate of change during the history of a lineage, nor is stasis precluded, but rather only abrupt or sudden change of great magnitude is precluded. In other words, by assuming change is "gradual," one is saying that time can be measured as a continuum rather than as discontinuous chunks.
p. 132      Seriation is a powerful ordinal scale method for ordering phenomena chronologically, because, as Dunnell (1981:67) points out, it can be applied in a wider variety of circumstances than can most other chronological methods.
p. 133      Variation in any of several properties of specimens (form and decoration, for example) provides an excellent basis for marking the passage of time (phyletic and evolutionary seriation) as do changing percentages of specimens placed in types (frequency seriation) or even the presence or absence of particular attributes or types (occurrence seriation). An ordering produced by seriation is just that, an ordering. Archaeologists hope that the ordering also represents a sequence, that is, a chronology, but this must be tested with other evidence independent of the seriation. But even a correct temporal sequence tells us nothing about the amount of time that elapsed during production of that sequence. We need an independent source of such information.
frequency and occurrence seriation p. 110      In short, the more types shared by two assemblages, the more similar they are, and thus the closer in time the two are thought to be; likewise, the more similar the frequencies of shared types of artifacts in two compared assemblages, the closer to one another the two are placed in an ordering and the closer in time they are thought to be. These comprise occurrence seriation and frequency seriation, respectively. Importantly, because the similarity of assemblages is measured in terms of shared types or type frequencies, the types must be theoretical units, or classes.
analogous vs homogenous p. 75      Thus we say that bird wings and bat wings are analogous structures, or analogues for short. They are of no utility in reconstructing lineages because they evolved independently in the two lineages after they diverged. The character of having wings is held in common by birds and bats, but the state of the character, the details of its osteological composition and anatomical structure, differs between the two groups. Analogous characters and character states are of no use in tracking heritable continuity but homologous characters, or homologues for short, are.
p. 83      Kroeber was suggesting that there are two forms of similarity: one homologous and the other analogous. The former results from shared genetic ancestry; the latter results from evolutionary convergence, such as when two genetically unrelated populations of organisms reach a similar adaptive solution.
superposition p. 144      Where units are superposed, or layered, the principle (some would say "law") of superposition states that the one on the bottom was deposited first and the one on top deposited last. Everything in between falls in relative chronological (depositional) order.
p. 145      We emphasize that the principle of superposition says nothing about the ages of the sediments - when they were formed - in each stratum relative to the ages of the sediments in other strata above or below it. It says only that one set of sediments was deposited at a particular locality before or after another set was. Conflating the age of sediments with the relative positioning of strata is the single-most common stratigraphical mistake made by archaeologists, and it can be a fatal one in terms of chronological ordering.
p. 173      Remember, the principle of superposition relates only to the chronological order of strata deposition, not to the age of the sediments or particles (including fossils and artifacts) making up the depositional layer.
stratigraphic excavation p. 149      We define stratigraphic excavation as removing artifacts and sediments from vertically discrete three-dimensional units of deposition and keeping those artifacts in sets based on their distinct vertical recovery proveniences for the purpose of measuring time.
p. 180      As we have noted and the heading of this section suggests, using superposition to measure time is an important part of an archaeologist's tool kit. But the adoption of stratigraphic excavation as the way to collect artifacts results in time being measured discontinuously, insofar as time is rendered as differences between superposed collections of artifacts.
index fossils p. 73      Rather, we are interested in creating an order of specimens that, when viewed overall, as in Fig. 3.3, measures the passage of time. This is about as close as we can come to measuring time continuously rather than as a series of chunks. If we pluck out a specimen from the continuum and turn it into an archetype, meaning that we use it as a representative specimen or average (norm) of numerous specimens, then we have created an index fossil.

Back to top: Michael J. O’Brien and R. Lee Lyman