Christopher Pierce

Paper presented at the 54th Annual Meeting of the Society for American Archaeology

Atlanta, Georgia 1989


Fire-altered rocks, as the name implies, are rocks that show signs of alteration resulting from exposure to extreme heat. These rocks are often referred to as "fire-cracked rocks" by archaeologists because cracking is a common and readily distinguishable form of alteration. The archaeological record in many areas of the world contains great quantities of fire-altered rocks and these rocks were recognized early on as artifacts. However, the lack of stylistic attributes of most fire-altered rocks lead, understandably, to their neglect by archaeologists devoted to working out culture histories. More recently, a growing concern with functional or analogous variation has lead some archaeologist, primarily in the past decade, to look more closely at fire-altered rocks as a source of information about the past (e.g., Ericson 1972; McDowell-Loudan 1983; Pierce 1982, 1988; Roll 1982; Thoms 1986; Van Dyke, et al 1980). Although these studies have served to increase the awareness of the potential of fire-altered rock studies in some areas, in general, archaeologists continue to ignore these artifacts. Consequently, we still know exceedingly little about the nature and significance of variation in fire-altered rocks.


My goal here is to demonstrate through two examples that fire-altered rocks possess considerable variation that is easily described and directly relevant to the kinds of functional questions asked by archaeologists today. In fact, as we will see, fire-altered rocks are ideally suited to functional studies because the class of artifacts is defined in terms that are directly related to a particular kind of use, that is the manipulation of heat energy. The examples will use are drawn from two different contexts. One includes data collected on rocks contained in six discrete concentrations or features exposed on the surface of a site. The other example focuses on fire-altered rocks recovered from test pits excavated in a shell midden where the rocks occur in varying amounts throughout the deposit rather than in discrete or recognizable features.




The six fire-altered rock features occurred in an eolian sand deposit in the Tiefort Basin located in the southeastern corner of the Fort Irwin Military Reservation in the Mohave Desert of southeastern California (Figure 1). In 1986, Far Western Anthropological Research Group conducted surface collections and limited excavations at this site (McGuire and Hall 1988). An attempt was made to collect all visible artifacts on the surface of the site including fire-altered rocks. All of the fire-altered rocks occurred in the six discrete clusters referred to on the map (Figure 2) as Features 1 through 6. 


Figure 1
Figure 1. Location of sites from which fire-altered rock collections were drawn.


The features range in surface area from 10 to 36 square meters and produced from 25 to 221 rocks. Excavations revealed intact subsurface deposits (lenses of charcoal and burned earth) in Features 1 and 6, and only a few small rocks in the surface levels of the other four features suggesting that these other features had been deflated (McGuire and Hall 1988). Only Feature 4, the largest of the six features, occurs with abundant quantities of other artifacts including chipped stone tools and debitage, ground stone and faunal remains.


Figure 2
Figure 2. Map of CA-SBr-5381.

Radiocarbon dates from three of the features range from 800 to 1000 years old. The rocks collected from the six features at Ca-SBr-5381 were identified as fire-altered artifacts on the basis of three criteria: 1) fractures indicative of thermal shock (70% of the rocks); 2) the occurrence of rocks in discrete clusters within an eolian deposit as evidence of their transport and deposition by people; and 3) association of rocks with other evidence of in situ burning.

The states of nine attributes were recorded on each of the 710 rocks recovered (Pierce 1988). In this paper, I explore the variation present in six of these attributes -- rock type or raw material, shape, size, fracture condition and fracture quality.


Rock Type 


We distinguished five rock types (basalt, granite, rhyolite, gneiss and quartzite) in the fire-altered rock assemblage from the six features at Ca-SBr-5381. This figure (Figure 3) shows the percentage of each rock type within the individual features. I should point out that this graph is not a frequency polygon and the lines connecting the data points are there to facilitate comparisons between the features. Although variation is evident between the features, there is no clear pattern except that gneiss is the most common rock type in most of the features.


Figure 3
Figure 3. Rock types present in fire-altered rock features.

The most significant aspect of rock type in the study of fire-altered rocks is that particular rock types react differently to heat because of differences in their chemical and structural compositions. Consequently, when we look at variation in the relative frequencies of different rock types, we must be certain that the rock types included do, in fact, have different thermal properties. This figure (Figure 4) shows experimentally derived values for several rock types of two common measures of the thermal properties of solids -- thermal inertia and thermal diffusivity. Thermal inertia is a measure of the rate of heat transfer at the surface of a substance. A rock type with a lower thermal inertia value is relatively less sensitive to temperature changes at its surface and is less likely to propagate heat to and from its interior than a rock type with a higher thermal inertia value. Thermal diffusivity is a measure of the relation between temperature change and the quantity of heat energy as it propagates through a substance. A rock type with a relatively low thermal diffusivity value produces less temperature increase than a rock type with a higher diffusivity value for the same amount of heat energy (Janza 1975:8ý-83).


Figure 4
Figure 4. Thermal properties of common rock types.

As you can see from this figure (Figure 4), basalt, granite and rhyolite have very similar thermal inertia and diffusivity values and, all else being equal, are virtually indistinguishable in their thermal properties. On the other hand, quartzite has considerably higher thermal inertia and diffusivity values relative to the other rock types present at Ca-SBr-5381. Although values for gneiss are absent from this data set, I suspect that the gneissic rocks from this site being highly siliceous and metamorphic are more similar to quartzite than to the igneous rocks.

When we group the three igneous rock types into one category and compare the relative frequencies of igneous, gneiss and quartzite rocks within the features, not only is there significant variation between featuresª but pattern emerges (see Figure 5). Features 1 and 5 have the greatest proportions of igneous rocks, Features 2, 3 and 6 are dominated by gneissic and quartzitic rocks and Feature 4 has a relatively even distribution across the three rock type categories. This variation between features is surely functional since the cost in terms of fuel consumption for producing a specific temperature change in the features dominated by igneous rocks would have been greater than in the features dominated by gneissic and quartzitic rocks. In addition, the features would have performed differently in terms of how long they held heat.


Figure 5
Figure 5. Fire-altered rock types grouped by thermal properties.



This graph (Figure 6) shows the relative frequencies of ovoid, blocky and tabular rocks within each feature. There appears to be very little variation between the features, but all of the features have far greater amounts of blocky rocks than other shapes. However, a contingency table analysis indicates that Feature 2 has significantly (p < .05) more tabular and less ovoid rocks and Feature 4 has significantly more ovoid rocks than expected under a model of random association. Even with these nonrandom associations, it seems unlikely that these differences would have significantly affected the overall thermal characteristics of the features. Although variation in shape can affect the thermal properties of rocks by altering the surface area to volume ratio, without more information on the distribution of rock shapes in the environment, it is difficult to determine the possible sources and significance of the fairly minor variation exhibited between the features.


Figure 6
Figure 6. Rock shape of stones in fire-altered rock features.



Rock size is an important attribute in two respects. Size affects the thermal properties of rocks by determining, along with shape, the surface area to volume ratio and the sheer volume of material involved. Variation in rock size may also indicate variation in the extent of rock use and re-useª since rocks fracture into smaller pieces as a result of repetitive exposure to heat. Observing the relative frequencies of different size classes of both uncracked and cracked rocks may provide a means of monitoring variation relevant to these two aspects of fire-altered rocks.

This figure (Figure 7) shows the distribution of uncracked and cracked gneissic rock sizes in the six features. Only gneissic rocks are used to control for difference in tenacity (resistance to breakage) of the various rock types present in the features. Feature 2 has a greater proportion of larger cracked and uncracked rocks perhaps indicating a functional difference in both selection and use of rocks in this feature. However, the general lack of difference in size distributions between uncracked and cracked rocks suggests that the uncracked rocks remaining are not a representative sample of sizes of originally selected rocks since size necessarily decreases with breakage. This pattern could be produced by smaller rocks being less susceptible to breakage through thermal shock and thus leaving the smaller rocks uncracked while breaking the larger rocks into smaller pieces. The larger uncracked rocks could also have been scavenged or recycled from these features. Only Feature 6 displays a marked difference in size between cracked and uncracked rocks possibly resulting from less extensive use of this feature. The predominance of rocks in the 0 to 5 cm size class might also indicate the size at which rocks became too small for re-use and therefore avoided further breakage, or may result from collection bias against smaller rocks.


Figure 7
Figure 7. Rock size for cracked and whole rocks in fire-altered rock features.



Fracture Condition


Another measure of the extent of use is the ratio of cracked to uncracked rocks. This figure (Figure 8) shows these data for gneissic rocks in the six features. The variation evident in this graph indicates that the features probably saw different amounts of use. Features 1 and 4 appear to have been more heavily used since they have considerably more cracked than uncracked rocks. Features 3, 5 and 6 have about the same amounts of cracked and uncracked rocks indicating less extensive use.  


Figure 8
Figure 8. Ratio of cracked to whole gneissic rocks in fire-altered rock features.


Fracture Quality

The kind or quality of fracture can also provide information on the use-history of fire-altered rocks. Physical principles and a growing body of experimental evidence suggest that spall-type fractures occur under different heating and cooling conditions than angular fractures (Adams and Waxler 1960; Blackwelder 1927; Draper and Stanfield n.d.; Roll 1982; Thoms 1986). However, several factors in addition to the heating and cooling regime can affect how rocks fracture under thermal stress (Pierce 1988). Most important among these are characteristics of the rocks themselves. To control for fracture differences due to rock type, this figure (Figure 9) includes only gneissic rocks.


Figure 9
Figure 9. Ratio of spall fractures of gneissic rocks in fire-altered rcok features.


Although angular fractures dominate in all features (reaching 100% in Features 3, 5 and 6), Feature 4 contains over five times the relative frequency of spall fractures in comparison to the other features. Recall on the previous graph that the pattern of variation between the features was quite a bit different than the pattern evident here. This suggests that the variation in fracture quality is not directly related to the extent of use of the features, but reflects differences in kind of use.




So far, we have focused on variation between discrete clusters or features of fire-altered rocks. However, fire-altered rocks often occur in more dispersed or continuous distributions without strong feature associations. What I would like to do now is quickly run through an analysis of fire-altered rocks from just such a context.


Figure 10
Figure 10. Location of CA-LAn-229 in the Santa Monica Mountains, California.

The site is a shell midden located in the central Santa Monica Mountains just west of Los Angeles (Figure 10). Surface collections and test excavations were conducted here in 1980 and 1981 by U. C. Santa Barbara (King et al. 1982). These investigations were restricted to a long, narrow transect across the site (Figure 11) and along which a series of small, 1 X .5 meter test pits were excavated. These excavations revealed a 30 to 50 cm thick midden covered by an equally thick layer of fine alluvial and eolian sediments generally lacking in situ cultural material. The presence of temporally diagnostic bead types indicates that the site was inhabited by people throughout the Late Period, or from approximately A.D. 1100 into the Mission Period (early 19th century).


Figure 11
Figure 11. Map of CA-LAn-229.

Because the physiographic setting precludes the deposition of sediments greater than 2.5 cm in diameter by geological processes (Pierce 1982), all rocks larger than this size were examined for evidence of modification. Almost 100% of these rocks, which are not chipped or ground stone artifacts display evidence of thermal alteration (cracking and discoloration). Sandstone and quartzite are the most common types of fire-altered rocks, and they occur in varying quantities throughout most of the deposit. Although some areas contain dense concentrations of rocks along with ash and burned earth, mixing by pocket gophers and the small size of the test pits made recognition of discrete features extremely difficult.


By plotting the ratio of the total amounts of sandstone to quartzite fire-altered rocks recovered from a string of test pits excavated in the central portion of the midden (Area 1), two areas containing radically different ratios are evident (Figure 12). The location of this deviation in the horizontal distributions of fire-altered rocks is matched by variations in other cultural materials including shell beads (King 1982:71-72) and subsistence remains (Johnson 1982:2, Graph 2; Bloomer 1982:23-25).


Figure 12
Figure 12. Horizontal distribution of ratio of sandstone to quartzite fire-altered rocks at CA-LAn-229.


To examine the vertical distributions of fire-altered rocks in these two areas, I averaged the ratios of sandstone to quartzite calculated for corresponding 10 cm levels excavated in the midden portion of the deposit in each area. This figure (Figure 13) shows the resulting distributions. If depth can be taken to equal time, it is apparent that the distinctive pattern seen in the horizontal distribution was not present when the site was first occupied, but a gradual divergence since that time has occurred. This pattern may indicate a change in the organization of the community.


Figure 13
Figure 13. Vertical distribution of ratio of sandstone to quartzite fire-altered rocks in Area 1 of CA-LAn-229.


These variations in the horizontal and vertical distributions of sandstone and quartzite fire-altered rocks cannot be accounted for by differential access to resources since both rock types occur in the same formation, which outcrops immediately north of the site (Pierce 1982). It seems more likely that the distinct thermal and mechanical properties of these two rock types lead to their selection for use in different activities.


At the outset, I stated that fire-altered rocks are ideally suited for functional studies and that this is true because the class of artifacts is defined in terms of use-related alteration. Because we know, by definition, that fire-altered rocks were used to manipulate the heat energy aspect of people's environment, we have been able to identify specific attributes of rocks that Necessarily affect their performance and cost when used in this manner. This knowledge has allowed us to talk about function and use without relying on untestable behavioral reconstructions. This is not to say that we are unconcerned with behavior.

Certainly, most of the patterns we have observed in the two examples are a product of human behavior. Rather, the difference lies in the use of physical properties of the record itself that do not change through time or space to gain access to behavioral variation in the past. In our examples, knowledge that the composition, size and shape of rocks affect their thermal properties and that exposure to heat causes rocks to fracture in certain ways made it possible to document spatial and perhaps temporal variation in function and use-history of fire-altered rocks occurring in discrete features and continuous distributions.

Finally, when we encounter fire-altered rocks during fieldwork, we generally record provenience and abundance data and then discard the rocks in the field. It is primarily for this reason that we know so little about the nature of variation in fire-altered rocks today. The two studies I have discussed here resulted from situations in which fire-altered rocks were collected and brought back to the laboratory for analysis. It is imperative that we begin to systematically collect and analyze fire-altered rocks as a normal part of archaeological investigations before opportunities to do so have vanished.



Adams, Leason H. and Roy M. Waxler
 1960 Temperature-Induced Stresses in Solids of Elementary Shape. Monograph 2. National Bureau of Standards, Washington, D.C.

Blackwelder, Eliot
 1927 Fire as an agent in Rock Weathering. Journal of Geology 35(2):134-140.


Bloomer, W. W.
 1982 Analysis of Shellfish Remains at LAn-229. In Archaeological Investigations at Talepop (LAn-229) compiled by C. D. King et al. Office of Public Archaeologyª University of Californiaª Santa Barbara. Submitted to the California Department of Parks and Recreation, Sacramento.

Draper, John, and Allan Stanfield
 n.d. The Cultural Significance of Fire-Cracked Rock at The Ozette Site: A Preliminary Study. Ms. in possession of author.

Ericson, J. E.
 1972 Geo-Science at the Castaic Site (4-LAn-324). Submitted to State of California Department of Parks and Recreation, Sacramento. Ms. in possesion of author.


Janza, F. J.

 1975 Interaction Mechanisms. In Manual of Remote Sensing, Volume 1, edited by F. J. Janza. American Society of Photogrammetry, Falls Church.


Johnson, J. R.
 1982 Analysis of Fish Remains from the Late Period Chumash Village of Talepop (LAn-229). In Archaeological Investigations at Talepop (LAn-229) compiled by C. D. King et al. Office of Public Archaeology, University of California, Santa Barbara. Submitted to the California Department of Parks and Recreation, Sacramento.


King, C. D.

 1982 Beads, Ornaments and other Artifcats Used to Maintain Social Relationships. In Archaeological Investigations at Talepop (LAn-229) compiled by C. D. King et al. Office of Public Archaeology, University of California, Santa Barbara. Submitted to California Department of Parks and Recreation, Sacramento.


King, C., W. Bloomer, B. Edberg, L. Gamble, J. Hammett, J. Johnson, T. Kemperman, C. Pierce, and E. Wohlgemuth 

 1982 Archaeological Investigations at Talepop (LAn-229). Office of Public Archaeology, University of California, Santa Barbara. Submitted to California Department of Parks and Recreation, Sacramento.


McDowell-Loudan, E.E.
 1983 Fire-cracked Rock: Preliminary Experiments to Determine its Nature and Significance in Archaeological Contexts. The Thesopiean, a Journal of North American Archaeology 21(1).


McGuire, K. R. and M. C. Hall
 1988 The Archaeology of Tiefort Basin, Fort Irwin, San Bernadino County, California. Far Western Anthropological Research Group, Davis, California. Submitted to the U.S. Army Corps of Engineers, Contract No. DACA-01-85-D-0100.


Pierce, Christopher
  1982 Stratigraphy, Features, and Site Formation. In Archaeological Investigations at Talepop (LAn-229), compiled by C. King et al. Office of Public Archaeology, University of California, Santa Barbara. Submitted to the California Department of Parks and Recreation, Sacramento.

 1988 Appendix H: A Functional Classification of Fire-Altered Rock Features from CA-SBr-5381. In The Archaeology of Tiefort Basin, Fort Irwin, San Bernadino County, California. Far Western Anthropological Research Group, Davis, California. Submitted to the U.S. Army Corps of Engineers, Contract No. DACA-01-85-D-0100.

Roll, Tom E.
 1982 Future Directions, Fire-Broken Rock. In Libby Additional Units and Regulating Dam Cultural Resources Project 1979, Descriptive Archaeology, edited by Tom E. Roll and Marc B. Smith. Montana State University, Bozeman.


Thoms, Alston V.
  1986 Hot Rocks: The Short-Term Storage of Energy in the Calispell Valley. In: Calispell Valley Archaeological Report, Interim Report for 1984 and 1985 Field Seasons, edited by Alston V. Thoms and Greg C. Burtchard. Center for Northwest Anthropology, Washington State University, Contributions in Cultural Resource Management No. 10, pp. 440-477.


Van Dyke, Allen P., David F. Overstreet, and James L. Theler
  1980 Archaeological Recovery at 11-RI-337, an Early Middle Woodland Shell Midden in East Moline Illinois. The Wisconsin Archaeologist 61(2):125-256.