The Lithic Containers
of the
Archaeological Record

Tony Baker
April 12, 2004

The Analogy
I spend a sizeable portion of my time each day walking for a break from my archaeological research. My walking environment is a nice residential neighborhood in metropolitan Denver. My environment has a major boulevard located about five blocks from my house. This boulevard contains all the popular fast food outlets. McDonalds and Taco Bell are within few blocks of my house and Wendys and KFC are located about 15 blocks away. Also, in my environment there are highly mobile critters. These critters have the unique behavior of driving up to one of these fast food outlets, purchasing a container of food with lots of napkins, condiments, etc., and then driving into my neighborhood. They only drive a few blocks before they locate a shade tree, under which they park. There they enjoy their meal. When they have satisfied their hunger, they put the remaining food, drink cups, napkins, and condiments into the food container and toss the container out of their window as they drive off. In my neighborhood, these critters are classified as Homo sapiens assholensis.

The Archaeological Record
One of my colleagues who read the above analogy asks, "what do assholes have to do with archaeology?" In this paper, the answer is nothing. This paper is about exhausted lithic containers, which are analogous to exhausted food containers. A lithic container is a core from which desired end products1 such as flakes, blades, or microblades, were removed by hard-hammer percussion. When the lithic container was exhausted, or emptied of its end products, it was discarded at or very near the quarry as was the food container discarded near the food outlet. When it was discarded a new chunk of lithic material was selected for reduction and the cycle began again. As a result, the density of exhausted lithic containers is highest at the quarry. This is the same density pattern exhibited by the food containers around the parent outlet. So by definition, a core made from exotic material is not an exhausted lithic container. To reiterate, an exhausted lithic container is a core of local material, which means it is found at the quarry, and it was reduced with hard-hammer percussion. An exhausted lithic container is not a tool or desired end product.

The archaeological record contains a number of different kinds of exhausted lithic containers, and this paper is concerned with three of them. The biface core is the oldest of the three, appearing around 1.5 million year ago. Subsequently, the Levallois core was developed circa 250,000 years BP and utilized until the end of the Middle Paleolithic (Gamble 1986:120; Garanger 1992:316). The Middle Paleolithic also saw the beginning of the blade core, but its presence became dominant in the Upper Paleolithic (Debenath and Dibble 1994:27). In contrast, during the Paleoindian and subsequent periods in the New World, the biface core remained the container of choice.
Table 1 -- End Product Characteristics
 Biface CoresClassic
Levallois Cores
Blade and
Sourceboth facesone faceone face
> 1.0> 1.0 and < 1.6 > 2.0

Similar to the sack and box food containers, these three lithic containers are different from each other. Table 1 lists some of these differences that are associated with the end products. End products from blade cores are indicated to be from only one face. The reader may question this because the summary literature seems to imply otherwise with the many pictures of conical blade cores. However, asymmetrical blade cores are the rule and not the exception. For example, from the Clovis Gault Site in Texas, there have been 105 blade cores recovered and only 14 are symmetrical, conical cores. The remainder are asymmetrical, acute-angle-platform cores with blade scars favoring only one face (Mike Collins, personal communication 2004). From Les Maitreaux, a Solutrean site in central France, there have been approximately 100 blade cores recovered and only one is a symmetrical conical core (Thierry Aubry,personal communication 2004). This is also true of the microblade cultures in the Arctic where the wedge shaped microblade core dominates the assemblages. Since, the majority of blade cores are asymmetrical; they are very similar to the Levallois core. Technically, the length-to-width ratio of the end product is really the only difference between the two. According to Van Peer (1992:41), "... the distinction between Levallois flakes and blades can only be a morphological one. Levallois blades are flakes with a length that at least doubles their width."

The biface core is different in that the end products were removed width-wise and from both faces. This is very important as will be seen later in the paper. That said, the three dimensions of length, width, and thickness will remain consistent throughout this paper. The longest dimension will be the length, the shortest is the thickness and the third dimension, which is perpendicular to both the length and thickness, is the width.

What Defines Exhaustion
When the last bite of hamburger is eaten and no more French fries remain, then the fast-food-container is exhausted. This is obvious. What is not so obvious is, why did the prehistoric knapper, who was working at the quarry, choose to discard a lithic container and start over with a new chunk of material? What was wrong with it? Why did he consider it exhausted? Did he abandon it for some cultural reason that is unknown to us, or did he encounter a physical constraint?

When I began this research, I suspected it was a physical constraint. This suspicion arose from my understanding of the behavior of modern knappers and my research on flake creation mechanics. Modern knappers do not create and thin bifaces for the end products as the prehistoric people did. They do it to make arrowheads from the center of the biface. Some of the more authentic modern knappers will begin their reduction of a chunk of lithic material with hard-hammer percussion, as did the prehistoric peoples. As the reduction proceeds and the biface becomes thinner, undesirable flakes in the form of steps and hinges start to develop. The modern knapper knows that when this stage is reached, he can either change to pressure flaking, or he can change to soft-hammer (antler) percussion. My observations of exhausted lithic containers in the archaeological record indicate that many of these have the scars of these undesirable flakes. So, I theorize that when the prehistoric knapper encountered this stage where undesirable flakes began to occur, he elected to start over with a new piece of material. This was in lieu of changing to pressure or soft-hammer percussion. Remember that the prehistoric knapper was reducing the lithic container for the end products and not some single tool that resided deep inside.

A physical constraint is predictable and, therefore, it can be discovered and measured. I chose to begin my search for this physical constraint with biface cores since I was most familiar with them and had personally acquired three data sets from different quarries. My first data set contains 13 biface cores from the Les Maitreaux site located in the Claise Valley of Central France. Les Maitreaux is a Solutrean, manufacturing site located in the Grand Pressigny flint quarry area. (See Aubry 2003 and/or Click on Les Maitreaux Site to view 12 of these 13 cores. My second set of data contains 13 cores from an Edwards quarry located in Crockett County, West Texas. The diagnostics at this quarry are dominated by Archaic cultures. The third set contains eight cores from the massive quarry area on the North Slope of the Brooks Range in Alaska. The core in the upper-right corner of the North Slope image is from the Mesa Site and the remaining cores are from various locations within 200 kilometers of the Mesa site. Their affiliations are unknown. (See Contrasting the Lithic Technologies of Mesa and Folsom). All 33 bifaces in the three images have been depicted at the same scale, and yes, the Les Maitreaux bifaces cores are much larger than the ones from the two North American quarries.2 To supplement these three data sets, I added 31 bifaces from the Paleolithic of Eastern Europe and Asia. I obtained the necessary data from the images in Paleolit S.S.S.R. (Boriskovskij 1984). The large bifaces from this source are classified as Acheulean handaxes. The disparity in size between the four data sets, plus the differences in time and space they represent, make them ideal for searching for a common physical constraint. If there is one, it must explain the variation in these data sets. Biface Core Data is a detailed listing of dimensions associated with these four data sets.

Figure 1 -- Thickness Model (TM) for Bifaces Cores
As discussed above, the modern knapper begins having problems reducing the biface with hard-hammer percussion when it becomes thin. Therefore, a physical constraint based on thinnest seems an appropriate place to start. Figure 1 is a plot of biface core maximum width versus maximum thickness for the four-biface core data sets. The linear regression line through the data is termed the Thickness Model (TM) in this paper. The regression line does not represent a constant width-to-thickness ratio because it does not pass through the origin. It is just the best-fit straight line with an R-squared value of 0.63. The R-squared value means that 63 percent of the variation in the widths of these 65 biface cores is explained by this straight line. For laboratory controlled experiments, an R-squared value of 0.63 is low and the TM would be considered a poor model to explain the variation in these biface cores. However, my experience with lithic data would suggest that 0.63 is not too bad. On the other hand, when one considers that the biface cores from Les Maitreaux, the Solutrean site, all plot above the line, this is not good. This TM appears to be culturally dependent. It does not represent a physical constraint that applies cross culturally. So if a physical constraint exists, it needs a different model.

Width-to-thickness ratio or the TM is related to biface flexibility, but it is not the true flexibility.3 A model that more closely approximates the flexibility of the biface can be derived from the flexibility of a cantilever beam. For a static loading condition, which is pressure flaking, width should be plotted against thickness times the cube root of length. When I plotted the data in this new model I obtained an R-square value of 0.74, which was an improvement. However, this model still suffered from the Solutrean data plotting on only one side of the line, indicating a cultural dependence.

Figure 2 -- Dynamically Loaded Model (DLM) for Bifaces Cores
Since one of my basic premises was hard-hammer percussion, I empirically derived a flexibility model for a cantilever beam under very rapid, dynamic loading. This model is termed the Dynamically Loaded Model (DLM) and its results applied to the biface core data are depicted in Figure 2. The DLM plots width versus the square root of the product of length times thickness. Now the linear regression line explains 87 percent of the variation in the biface width and the cultural dependency associated with the Solutrean bifaces has vanished. All four data sets plot on both sides of the line.

The DLM suggests that the average biface core became exhausted and was abandoned when its flexibility increased to a fixed value defined by the regression line in Figure 2. Bifaces that plot below the line are stiffer than those on the line. Bifaces that plot above the line are more flexible. Is it possible that the other two lithic containers, Levallois cores and blade cores were abandoned for a flexibility reason?

To test Levallois cores against DLM and compare it to the TM, I found two sources of data. First, the largest and most extensive source was Table 1 in the Appendix of The Levallois Reduction Strategy (Van Peer 1992:122). This Table lists average metric values for cores from 35 Africa Levallois sites. These 35 sites represent 1350 classic Levallois Cores, ranging from 10 cores in the smallest site to 96 cores in the largest.

Figure 3 -- Thickness Model (TM) for Levallois Cores
My second source of Levallois data comes from the prehistory collection of the British Museum. While on holiday in January 2004, I spent a few hours analyzing some cores from Baker's Hole, Northfleet. This site has yielded some extremely large cores as can be seen from the data plotted in Figure 3. Roe (1981:214) refers to the extremely large end products from Baker's Hole as "bold flakes". He (1981:215) further writes, "Baker's Hole is in many ways a unique site; the relatively early date suggested for it (inter-Wolstonian) may have something to do with this, but the abundance, high quality and large size of the flint nodules available there are probably more powerful factors." The staff at the British Museum selected these seven cores for me to study and they consumed the time I had allotted. I cannot say if they represent a random, statistical sample or not. Levallois Core Data details the data from Table 1 of Van Peer's book along with the cores from Baker's Hole.

Levallois cores are mechanically different from biface cores. As stated in Table 1, end products were removed lengthwise from the cores. This is in contrast to the width-wise removal from biface cores. The direction of end product removal is important to the geometry of both the TM and the DLM. The length and width must be transposed in the models for Levallois cores. For the TM in Figure 3, length is plotted against thickness. For the DLM in Figure 4, length is plotted against the square root of the product of width times thickness.

The first thing one notices about these two models are how well each explains the variation in length of the Levallois cores. The DLM is slightly better, with 97% of the variation being explained as compared to the 92% for the TM. These are exceptionally high values. Most likely they are largely a result of the Van Peer data being average site values, which removes the variation between individual artifacts, leaving only the variation between sites. The ability of both the TM and the DLM to explain the variation almost equally well is more perplexing. However, when one realizes that with Levallois cores, the length and width are highly correlated, it is mathematically explainable. This high correlation makes the DLM and the TM mathematically the same.4 This is independent support for the DLM being the correct model.
Figure 4 -- Dynamically Loaded Model (DLM) for Levallois Cores

Blade cores provide additional insight into the concept of lithic containers being abandoned because of a physical constraint. Additionally, they offer a third type of lithic container that can be used to test the DLM. In this paper, blade cores include both big-blade (macroblade) and microblade cores. All morphologies are also included, such as conical cores with almost 90 degree platform angles, prismatic cores with acute platform angles, and wedge-shaped (boat-shaped) cores. The data I used are a mixture from the literature and cores I personally measured. Table 2 lists the literature sources and the number of cores I obtained from each. It also lists the 79 cores I personally measured. As can be seen, American Beginnings provided most of the data and most of these are wedge-shaped microblade cores that cluster at the low size end of Figures 5 and 6. Of the cores I personally measured, 74 were large cores and are from the Clovis level of the Gault Site in Texas. Blade Core Data provides the detailed dimensions of all 208 cores.
Table 2 -- Source of Blade Core Data
Source# of
American Beginnings (West 1996)106
Clovis Blade Technology (Collins 1999)
only non-Gault Site cores
Hengistbury Head Dorset Vol 2 (Barton 1992)8
Parrain Nord (Gaussen, Hesault, Joyel 1994)2
Cores the author personally measured,
mostly from the Gault Site

Figure 5 -- Thickness Model (TM) for Blade Cores
Figure 6 -- Dynamically Loaded Model (DLM) for Blade Cores
Reviewing the TM (Figure 5) and DLM (Figure 6) for blade cores, it is evident that the DLM is again the superior model, explaining 79% of the variation of the blade core length in contrast to 73% for the TM. The R-squared value for the DLM is similar to the value for the biface cores (Figures 2), but less than that for the Levallois cores. Again, I believe the Levallois value is too high because it is derived from average site values and not the individual artifacts. The R-squared value for the TM is close to the DLM value as it was with the Levallois cores and for the same reason.

Figure 6 has a second story to tell. The vast majority of cores from Beringia (Alaska and Siberia) plot near the lower end of the regression line. In contrast the cores from Texas, of which 74 of the 87 are from the Clovis level at the Gault Site, plot in the middle to upper portion of the regression line. In fact, the shortest Texas core is 66 millimeters, which is longer than 82% of the Beringia cores. This obvious difference is due to the Beringia cores being a mixture of many microblade cores and a few big blade cores, while the Texas cores are only big blade cores. In different words, the desired end products were different for the two data sets. Beringia had a microblade-inset technology and Texas did not. Since microblade cores could have been made from the Texas lithic packages, which are obviously sufficiently large, this is a cultural constraint. I will have more to say on small lithic packages in the next section.

Summary and Discussion
This paper began with a comparison of the lithic container to the fast food container. A lithic container is defined as a core (biface, Levallois, or blade) from which end products (flakes and blades) are removed by hard-hammer percussion. When the lithic container was exhausted, it was discarded at or near the quarry, as is the fast food container tossed on the streets near the fast food outlet. Subsequent to this opening analogy, I proceeded to define exhaustion of a lithic container as a physical constraint to further reduction with hard-hammer percussion. This physical constraint is a result of the lithic container becoming too flexible and, therefore, undesirable end products start to occur.

I used portions of the archaeological record from around the World and from the Lower Paleolithic to the recent Archaic times to define this physical constraint. I tested a number of models before I discovered the Dynamically Loaded Model (DLM), which I initially, empirically derived from a computer program describing the response of a cantilever beam to dynamic loading. The DLM, which is represented by the regression lines in Figures 2, 4, and 6 for the three types of containers, is the best model I have found based on the amount of variation it explains. These regression lines are different because each lithic container type has a different cross-section relative to the direction of end product removal. However, they each take the form of y = mx + b; with y being the dimension parallel to the direction of the end product removal, and x being the square root of the product of the thickness and the dimension perpendicular to the direction of the end product removal. Refer back to Table 1. The DLM was derived from dynamic loading conditions, because pressure flaking (static loading) does not place a physical constraint on thinness. In different words, hinge flakes are not a product of pressure flaking. The thinness constraint on pressure flaking is the knapper's skill.

Figure 7 -- Idealized Exhaustion Line
To better understand the archaeological record at quarries, I have created Figure 7, which depicts an idealized exhaustion (limiting flexibility) line. Lithic containers that plot to the right or below the exhaustion line are not flexible enough to be exhausted. So, the containers that plot at Locations 1 and 3 are not exhausted and they contain additional end products available to be removed by hard-hammer percussion. Containers that plot on the line, which are similar to Locations 2 and 4 are at the limiting flexibility (exhausted) and hard-hammer percussion will not yield additional desirable end products.

Let's assume that the life of a lithic container begins at Location 1. End products are removed by thinning and the lithic container slowly moves toward Location 2. This is horizontal movement because the dimension, which is parallel to the end product removal direction, is not being shortened. When Location 2 is reached, undesirable end products with hinge and step terminations start to form and the lithic container is exhausted. A different scenario often happens with a biface core. During the transition from Location 1 to 2, the biface will often break into two pieces. This instantly reduces the length but not the width, which causes the separate pieces of the original biface to move to locations similar to 2a. These new locations represents flexibility greater than the exhaustion line and I call this super-exhaustion. The biface was not near exhaustion prior to the break, but the breaking instantly exhausted it and the two pieces are discarded. This is analogous to a hole appearing in one's food container and the contents instantly spilling on to the ground. In the archaeological record, broken biface cores are extremely common at the quarries and I argue they are there because of the above scenario and not because they were tools that were lost in various stages of manufacture.

The archaeological record associated with blades tells another interesting story. Many blade cores from quarries exhibit step and hinge scars from blades that originated from platforms that no longer exist. In different words, these undesirable blade scars were created when the core was longer. Referring back to Location 2 in Figure 7. At this location the blade core is exhausted and undesirable end products are being produced. If the knapper should remove a core tablet, he shortens the core, which stiffens it, and moves it to Location 3. Now the blade core is no longer exhausted and more desirable end products can be removed. Additional removal of blades will move the core from Location 3 to location 4. At Location 4 the blade core can again be shortened by removing a core tablet and then additional blades can be removed. If this recycling process continues, the blade core becomes a microblade core. If the recycling stops along the way, as in the case with the Texas blade cores, then the knapper obviously desires a certain length blade. This is a culture constraint. The concept of recycling blade cores by shortening them can also apply to Levallois cores and there is some evidence it was employed (Van Peer 1992:15-33).

Many artifacts in the archaeological record plot in the super-exhaustion region (left of exhaustion line) of Figure 7. However, they are in this region because of a technology different than hard-hammer percussion. Soft-hammer (antler) percussion will plot there, but it too has an exhaustion line or physical constraint. It is just more to the left than the hard-hammer percussion line. Pressure flaked artifacts can plot to the left of this line, and as mentioned above pressure does not have a physical constraint due to thinness. The distance into the super-exhaustion region depends on the skill of the knapper. Both soft-hammer percussion and pressure usually occur away from the quarry when the knapper is trying to extend the useful life of lithic material. The classic example of this behavior is Folsom technology.

Figure 8 -- Composite of the Lithic Containers' Data
A second physical constraint to hard-hammer percussion also exists. Figure 8 is a composite of the three lithic containers used to develop Figures 2, 4, and 6. Notice there are no lithic containers below 20 millimeters on the vertical and 12 millimeters on the horizontal. These unpopulated regions defined by these minimums represents this second physical constraint. Hypothetical lithic containers that might occur in these regions have so little mass that the inertia effects required for hand held percussion knapping is not sufficient to permit end-product removal. They are just too small. Also, notice how all the data comes together at the small end. All three lithic containers types overlay each other. The cross-sectional geometry of the lithic container is no longer a factor and only mass is important. All the regression lines would probably pass through the origin if larger data sets had been employed.

At the large end of the lithic containers, which is the upper right of Figure 8, the exhaustion lines separate. For example at 80 millimeters on the horizontal axis, the vertical values for Levallois, blade, and biface exhausted cores are 157, 130, and 85 millimeters, respectively. The horizontal value of 80mm is the square root of the product of the other two axes and, therefore, the volume of these lithic containers can be calculated. At 80 millimeters on the horizontal, these volumes are 1010000, 830000, and 540000 cubic millimeters, respectively. Biface cores waste the least amount of material when discarded. Levallois wastes the most. Notice, I have said nothing about the volume of desired end products removed before exhaustion. At this writing, I have no way to assess this.

In closing, I want to suggest that the next time the reader is walking through a quarry site and you pick up a cool tool. Ask yourself, is this really a cool tool or just an exhausted lithic container?


1 "End product" is a collective term use herein to describe the desired flake, blade, or microblade that was removed from a lithic container. I borrowed the term from Old World Archaeology, where it is generally associated with only the desired flakes and blades of the Levallois technology.

2 Les Maitreaux is located in the Grand Pressigny area where the lithic material comes in extremely large packages. Therefore, it is possible to make large bifaces cores. The large piece of material in the image at the top of this paper was found 30 meters from the excavation at Les Maitreaux. The material is so abundant in the area that the Solutrean peoples never utilized this huge piece. In fact, it was not even used by the subsequent people that surely must have seen it.

3 "Flexibility" is used here to mean the opposite of "stiffness." Something that is flexible is not very stiff.

4 The Dynamic Loading Model for Levallois cores has the form of a straight line or:
 Eq. 1
Most likely the DLM passes through the origin because in the real world, if either the width or thickness is zero, then the length is also zero. Although the limited data I used produced a regression line that did not pass through the origin, I suspect that if additional data were available it would do so. Therefore, if I assume that the intercept (b) is zero, Eq. 1 becomes:
Eq. 2
Squaring both sides of the Eq. 2 we have:
Eq. 3
The correlation factor between width and length for the Levallois core data used in this paper is 0.98. In comparison, the same correlation factor for the biface data is 0.64. Since the width and length for the Levallois cores are so highly correlated, and again there is probably a zero intercept because if the length is zero, then the width is zero: I can write for Levallois cores:
Eq. 4
Combining Eq. 3 and Eq. 4:
Eq. 5
Dividing both sides by length and combining the constants into C, the equation becomes the Thickness Model with a zero intercept:
Eq. 6


Besides the usual people who are listed on my home page, I want to thank Nick Ashton, Alan Slade, and Debbie Slatford of the British Museum for permitting me to study and measure a number of the Levallois cores in the Museum's collection. They were helpful and gracious on very short notice. Second I want to thank Thierry Aubry and Mike Collins for permitting me to do the same with the bifaces cores from Les Maitreaux and the blade cores from Gault. Finally, I want to thank Natalia Slobodina for her English translation of certain parts of Paleolit S.S.S.R..


Aubry, Thierry, M. Almeida, M. J. Neves and B. Walter.
2003  Solutrean Laurel Leaf Point Production and Raw Material Procurement During The Last Glacial Maximum in Southern Europe: Two examples from Central France and Portugal. In Multiple Approaches To The Study of Bifacial Technologies. M. Soressi and H.L. Dibble, ed. Pp. 165-182. University of Pennsylvania Museum of Archaeology and Anthropology (University Museum Monograph 115), Philadelphia.
Barton, R.N.E.
1992  Hengistbury Head Dorset -- Volume 2: The Late Upper Palaeolithic & Early Mesolithic Sites. Oxford University Committee for Archaeology, Oxford.
Boriskovskii, P. I.
1984Paleolit S.S.S.R. Nauka, Moscow.
Collins, Michael B.
1999Clovis Blade Technology. University of Texas, Austin.
Debenath, Andre, and H. Dibble.
1994The Handbook of Paleolithic Typology. Vol. I. The Lower and Middle Paleolithic of Europe. University of Museum Press, Philadelphia.
Gamble, Clive
1986The Palaeolithic Settlement of Europe. Cambridge University Press, Cambridge.
Garanger, Jose
1992La Préhistoire Dans Le Monde. Presses Universitaires de France, Paris.
Gaussen, J., J.-C. Hesault and S. Joyel
1994Parrain Nord Station Magdalenienne de Plein Air. Paleo: Revue d'archeologie prehistorique La Societe des Amis du Musee National de Prehistoire et de la Recherche Archeologique, Les Eyzies de Tayac.
Roe, Derek A.
1981The Lower and Middle Palaeolithic Periods in Britain. Routledge & Kegan Paul, London.
Van Peer, Phillip
1992The Levallois Reduction Strategy. Prehistory Press (Monographs in World Archaeology 13), Madison.
West, Frederick H.
1996American Beginnings -- The Prehistory and Palaeoecology of Beringia. The University of Chicago Press, Chicago.

Return to Paleoindian & Other Archaeological Stuff

eXTReMe Tracker