Thermal fracture

Pictured below are two parts of the same lithic broken by a frost-fracture (the piece on the left has been turned upside down). The irregular nature of the break is very characteristic of this type of fracture, caused by changes in temperature and often originating from existing "faults" in the flint matrix. The distinctive features of this fracture are; both parts have been recovered in-situ with minimal separation (a few grains of sand separating them), there is a marked difference in the patination from that of all the other planes which as a group are contiguous, the fracture plane is irregular - characterised by angular planes (often concave) and finally, the fracture appears to originate from within the centre of the flint. Typically flint broken by thermal fracture is irregular in nature.

Artefact broken in half by frost fracture

Mechanical fracture

In the next photograph the same artefact can be seen pieced together with the distal end of the original fracture surface of this fairly large and thick primary flake facing upwards. The bulb of percussion featured here is pronounced, as are the ripples where the force of impact travelled. Further reductions (by direct percussion and not thermal-fracture) have removed both the tip of the distal end and the initial platform on the proximal end. There has been extensive unifacial retouch to both lateral edges, including the manufacture of a robust burin point. Using lithic classification this is a multi-purpose tool with all major fracture planes caused by mechanical force or "direct percussion" (there are additionally many small micro-fractures caused by the pressure of the matrix over very long periods of time).

Determining the difference between a natural and cultural assemblage: A repeatable analysis attempting to counter the affect of bias.

Introduction

A common charge, and not one only made in reference to this assemblage, is that the items displayed are fortuitous, derive from natural processes and do not appear to be the result of human agency. But these "opinions" are countered in this case at least, by others that note "some agency" or "definite workmanship", for example; by two independent and eminent geologists remarking that several items were specifically "not the result of natural processes".

Gillespie et al (2004) provide one of only a handful of papers that address the issue of detecting human agency objectively. Subjective views abound but are of little use to a science that demands repeatable methodology. Their paper remarkably treads more or less un chartered terroritories by identifying (sixteen) statistically significant attributes to determine objectively* whether an assemblage is natural or cultural. SInce this assemblage attracts more than perhaps its' fair share of challenges with respect to being authentic ancient artefacts, it seemed only appropriate to apply this test and be open and honest about the results.

I found an unlikely collaborator overseas who happily did not subscribe to the opinion that the assemblage was cultural. We talked, even argued about item after item, but rarely found common ground. In the end I put this test to him as a fair way of determining one way or another - subjectively and without recourse to either of our biases, that were, in the course of our discussions, often more than apparent. The artefacts would stand the test or fail on their own merits. In the process, we found a workable method of assessing photographs of lithics over the internet and agreement.

After some deliberation, and certain caveats the challenge was agreed and accepted, but no commitment to publication was given.

"The belief that anyone can select a lithic industry of any sort given a big enough rock pile has never been demonstrated. I have challenged many to try, imposing the limitation that the selection must be from time levels at which there is no possibility of man's presence. No one has been able to do so. No artifact-like material, no patterned lithic industry emerges, even in areas of easily flaked rock. This is a myth comparable to that of the infinite number of monkeys with typewriters reproducing Shakespeare's work."

George Carter in response to the critics of Raemsch's and Vernon's research (Timlin site).

One hundred items would be collated that were believed (by me) to be the result of human agency (sampling methodology described below). From this sample fifty would be selected at random for analysis. The interpretation of each attribute (described later) and the methodology of scoring would be ratified by both parties which was fairly quickly and easily achieved on mutual terms. Although it was recognised by both parties that the attributes may have been improved, or even adjusted for the proposed "cultural type" of material, the test would be conducted as set out in the original paper in order to provide a comparable method and data set (i.e. the artefact set, the naturefact set, and the original sample set - which was subsequently determined to be natural).

* Ultimately some of the attributes are some-what subjective, but when the results are a) formed from more than one independent opinion (with group deliberations on those results that differed) and b) considered as part of the collective score, this to some extent counters latent subjectivity or bias.

Method

The random sample was initially selected from boxes of lithics collated from my initial excavation, but not deemed of any immediate interest. I simply ensured that I did not include items that were not "worked" - many of which I had kept. After collecting up what I felt were enough in several trips from my shed, I proceeded to count through them and ensure that I had not selected any flakes (which we had agreed would not be included in the experiment). I had just shy of one hundred, so I randomly selected some from a box indoors of 'iconic' material to make up the numbers. Each item was numbered from one through to one hundred in no particular order. Selection numbers were generated randomly by my colleague and provided after I had numbered, photographed and emailed the pictures to him. I sighed as some fairly "safe" ones were selected because I was beginning to worry I was going to lose this test and grinned when a couple of the 'iconic' ones were too. This was looking like it could be interesting.

Each item was then parsed through one attribute at a time to ensure that a consistent approach was made on each occasion, the initial judgement was of course mine (who else would have found time? I even advertised for independent assessment). The results for each item and attribute are recorded and summarised at the end of this page. When complete I took photographs of a few items to begin with and discussed the results with my colleague. We went through each attribute one at a time, and then finally compared our joint analysis (i.e. agreed) and scored accordingly. Essentially we also compared my original scoring with the agreed scoring. There was very little difference from the examples we assessed together and in the process I feel I learnt a little more again about lithics. However, it soon became apparent that the overall result was not going to change markedly and for this reason we regrettably did not progress through the majority of the sample together. The data shown therefore is a combination of my initial guided assessment and joint collaboration on only a few items. However, I am more confident now than I was at the beginning of the judgement on each occasion, the results are public, and the collection is still available for further analysis.

Results

Figure 1 is a summary graph depicting the results (in red) in comparison with the results from the Gillespie paper.

Figure 1. Score for Test Sample in comparison to original data from Gillespie 2004.

N.B. T1* is the name I have given to the original sample tested in Gillespie 2004 to differentiate it from my test sample.

Conclusion

The results indicate that the "test sample" was far closer to the artifact sample than either the geofact set or the original Gillespie sample subsequently identified as non-cultural. Therefore I conclude that even if the results were reduced somewhat by providing counter arguments for some attribute scorings, the sample is unlikely to consist of a majority of geofacts.

However, I recognise that there are weaknesses with this experiment, not least of which being that I did not judge the original samples, and Gillespie et al did not judge these. Some may argue, with some ground, that the results are therefore not directly comparable. But, since the method had been set out fairly clearly - indeed as a good scientific methodology should - and the interpretations of these definitions were arrived at independently and in joint agreement, the standard of assessment is at least consistent and therefore the results cannot be dismissed in their entirety. The methodology of analysis is repeatable, as we showed, and the attributes were determined to be reliable indicators of agency when taken as a whole. Indeed, if the test was to be adapted to be more sensitive to an archaic assemblage where unifacial working is more frequently encountered for instance then perhaps the assemblage would have faired even better? Another attribute that may also be adapted is that with relation to cortex since complete removal could be said to be uncommon in some lithic industries.

For the moment the evidence appears to rest in favour of the assemblage being cultural, and this is where it will remain until proven otherwise and not by "appeals to authority" as is often the case in this field.

Table 1. below shows the culmulative score for the test sample. All other data is taken from Gillespie 2004.

The results for the test sample are provided in Table 2 below.

Table 2. Attributes applied as detailed in Gillespie 2004 and detailed below.

"1. Platform Preparation

During the manufacture of cultural artifacts, the knapper attempts to gain maximum control over flake removal by manipulating the dimensions and angle of the striking platform. Although the platform is usually removed during reduction, a remnant may remain. The presence of platform preparation was recorded when all remnant platforms were cortex-free and exhibited small micro-flaking at the location of flake removal, suggesting intentional platform preparation (Patterson, 1983:302).

2. Uniface/Biface

While both natural and cultural processes can produce bifaces Schnurrenberger
and Bryan, 1985; Chlachula and Le Blanc, 1996), it was hypothesized that bifacial reduction would be more common in culturally produced assemblages. This is particularly true, given that many of the known natural cobbles had only a few flake scars (Table I). While a tool of human-manufacture may be unifacial, it is hypothesized that bifacial reduction is more common. This type of statistical argument has been used for many of the lithic attributes. While exceptions may be found, the utility of each attribute is based on whether it statistically differentiates artifacts from geofacts. Thus, if any part of the cobble was flaked on both faces, it was awarded one point.

3. Differential Weathering

If a cobble has been flaked by natural processes, one would expect a significant period of time between flake removals. This may be manifested as differing degrees of weathering on flake scars from the same cobble (Oakley, 1959:19). If a cobble did not exhibit noticeably different degrees of weathering on separate flake scars, it was awarded one point.

4. Inverse Bulb

When percussion is used to remove a flake from a cobble, an inverse bulb may be left on the core surface. Because a significant degree of force must be applied at the proper angle to create a bulb of percussion (and therefore an inverse bulb), this attribute is more likely to exist on artifacts than on geofacts. It has been argued that natural flake removals produce either a shallow diffuse bulb of percussion or, frequently, no bulb at all (Peacock, 1991). The presence of one or more obvious inverse bulbs on a cobble scored one point.

5. Percentage of Cortex Present

While the amount of cortex present on a cobble will vary widely, increased modification results in less cortex. Therefore, a human-produced cobble may have less cortex than a geofact, especially when the cobble is intended to be a tool rather than a core. Obviously, some tools will still have cortex, but it is unlikely that a geofact would be entirely free of cortex. Conservatively, a cobble was awarded one point if it had no remaining cortex.

6. Scar Alignment

The argument has often been made that human-manufacture produces a “logical” flaking pattern, while natural processes result in random or “illogical” flaking (Patterson, 1983; Chlachula and Le Blanc, 1996). This has been difficult to characterize in a systematic fashion. It was thought that scar alignment might test this hypothesis. Parallel flake scars, especially if adjacent and/or overlapping, would be indicative of human-manufacture. Random flake-scar alignment would be indicative of natural-manufacture. Only large, primary flake scars were used in this category, not secondary retouch scars. One point was awarded to a cobble if all primary flake scars were parallel.

7. Secondary Retouch

A human-manufactured cobble may exhibit secondary retouch (Chlachula and Le Blanc, 1996). Secondary retouch is defined as the presence of additional flake removals from a previously flaked edge of a cobble, usually to create a better working edge or for platform preparation. Secondary retouch flakes are smaller than primary flakes. This attribute was recorded when a cobble showed any indication of modification subsequent to primary flake removal. We chose to remain conservative, so crushed or ground edges were not recorded as secondary retouch, as this process could occur in both natural and cultural settings. One point was awarded when secondary retouch was present.

8.Size Uniformity of Edge Flake Scars

This is also an extension of the retouch attribute. It was hypothesized that human manufactured retouch would produce flake scars of a relatively uniform size. Natural edge modification would produce flakes of varying sizes, depending on the type and extent of force during removal. Obviously, even on a cultural cobble, all flake scars will not be the same size. Therefore, a small amount of variability was permitted. The degree of variability accepted was judged on a cobble-by-cobble basis and based on the overall flaking pattern. Interestingly, although this attribute is subjective, the chi-square results suggest that, if fairly applied, it can differentiate artifacts from geofacts. One point was awarded when edge modification consisted of flake removals of a uniform size.

9. Arrises

Flakes removed during human-manufacture will exhibit deep negative scars, while natural flake removals often result in shallow flake scars (Chlachula and Le Blanc, 1996). This should result in well-defined arrises (i.e., the edge representing the border of the flake scar). Furthermore, if a cobble has been subjected to substantial post-depositional alterations, arrises may be worn. Consequently, it is hypothesized that well-defined arrises would be more indicative of artifacts. Small changes in angle, speed, and degree of force can alter the shape and size of flake scars. As a result, there is no absolute cutoff point separating well-defined from diffuse arrises. Each cobble was scored by comparing it to the other assemblages. One point was awarded when all arrises were clear and well defined.

10. Low-Angle, Alternate Biface Flaking

While the presence of any type of bifacial flaking appears to be indicative of artifacts, low-angle, alternate bifacial flaking is believed to be particularly indicative of human-manufacture (Chlachula and Le Blanc, 1996). This attribute was recorded when the cobble had an edge that resulted from several flake removals on both faces. The edge also had to result from alternating flake removals (i.e., one removal from the first face followed by one from the second) and the end result must have been a low angle (< 75°). While this may prevent some core tools from scoring in this category (edges near 90°), they would not be eliminated from other categories, as each attribute is independently scored. One point was awarded when bifacial flake removal alternated, producing a low-angle edge.

11. Striations on Tools

The next four attributes attempt to measure the effect of context rather than the cobbles’ manufacture. While both artifacts and geofacts can experience post-depositional alterations, we hypothesized that geofacts would experience greater postflaking alteration than artifacts, given the context of the assemblages used in this study. This is based, in part, on Chlachula’s (1994a) claim that only low-energy processes were available at the proposed pre-Clovis sites. If this interpretation is correct, little post-depositional alteration would be expected. On the other hand, geofacts are nothing more than clasts within the till deposit and are subject to continued taphonomic alternation (i.e., load and shear stress). Therefore, striations anywhere on the cobble may be more common on geofacts. Finally, striations may be indicative of shear stress, which may suggest flaking in an active glacial environment. Conservatively, one point was awarded if the cobble had no striations, indicating no post-flaking alteration.

12. Striations on Flake Scars

Striations on flake scars should be even more indicative of post-depositional alteration because they must have occurred after reduction. A cobble was awarded one point if no striations were found on any flake scars. While using striations on both cobble and the flake scars results in “double-counting” of a similar attribute, it was thought that by separating these similar features, it would be possible to differentiate pre- and post-depositional alterations. Furthermore, although double counting may seem to give too much weight to a given set of attributes, it is important to remember that all cobbles were scored using the same criteria. In other words, even if two attributes record a similar phenomenon, it does not bias the results, as each assemblage was treated equally.

13. Pecking on Tools

Like striations, pecking would indicate post-depositional alteration. Pecking results in small impact scars on the cobble surface. However, unlike striations, pecking would indicate the presence of impact force common to high-energy fluvial and colluvial environments. Again, this attribute provides a general record of past taphonomic processes. One point was awarded if no pecking was found on the cobble.

14. Pecking on Flake Scars

We hypothesized that pecking on flake scars would be clearly indicative of postflaking alteration. One point was awarded if no pecking on flake scars was found. The previous four attributes provide some indication of context, and thus are only useful in analyzing the Alberta assemblages. They were designed to test specific claims about the geological context at the proposed pre-Clovis sites in question (Chlachula, 1994a; Young et al., 1998; Osborn et al., 2000). Sites in other geological contexts would require different attributes.

15. Logical Flaking

This is a subjective attribute that incorporates all other attributes. Logical flaking was recorded when the cobble appeared to have been reduced in such a way that a usable edge was produced, or flakes were removed in an efficient manner (e.g., using previous scars for subsequent removals). Obviously, this assumes that our interpretation of “logical” is similar to that of the prehistoric knapper. This attribute was included because many researchers have used it in their analyses (e.g., Patterson, 1983; Chlachula and Le Blanc, 1996). By itself, it would provide little interpretive value, but because it is only one of 16 attributes used in this study, it does not unduly bias the results. Furthermore, Table I shows that logical flaking was recorded on several of the known natural and proposed pre-Clovis cobbles. One point was awarded when logical flaking was present.

16. Number of Flake Removals

While the number of flake scars is dependent on many factors (e.g., lithology, tool use, production method), it was hypothesized that artifacts would have more flake removals than natural cobbles. A cobble was awarded one point if it had more than five flake scars. The cutoff point of five flake scars was chosen because it seemed a reasonable number in differentiating natural from cultural cobbles. In fact, most of the known natural cobbles had less than five flakes while most of the artifacts had more than five. The proposed pre-Clovis assemblage had an average of 3.5 flake scars.”