University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Anthropology Faculty Publications Anthropology, Department of

1988

On the Nature and Antiquity of the Manix Lake Industry

Douglas Bamforth University of Nebraska - Lincoln

Ronald Dorn Arizona State Univ., Tempe, AZ

Follow this and additional works at: https://digitalcommons.unl.edu/anthropologyfacpub

Part of the Anthropology Commons

Bamforth, Douglas and Dorn, Ronald, "On the Nature and Antiquity of the Manix Lake Industry" (1988). Anthropology Faculty Publications. 24. https://digitalcommons.unl.edu/anthropologyfacpub/24

This Article is brought to you for free and open access by the Anthropology, Department of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Anthropology Faculty Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Journal of California and Great Basin Anthropology Vol. 10, No. 2, pp. 209-226 (1988). On the Nature and Antiquity of the Manix Lake Industry

DOUGLAS B. BAMFORTH, Dept. of Anthropology, Univ. of Nebraska, Lincoln, NE 68588-0368. RONALD I. DORN, Dept. of Geography, Arizona State Univ., Tempe, AZ 85287.

X HE antiquity of human occupation in the THE MANIX LAKE INDUSTRY New World undoubtedly is one of the major Prehistoric Manix Lake existed in what is unresolved culture-historical problems in now the valley of the Mojave River east of North American prehistory. On the one Barstow, California (Fig. 1). Although this hand, a dominant position with a long his­ region is now extremely arid, geomorphic tory in American archaeology (cf. Wilmsen and stratigraphic evidence clearly shows that 1965) holds that human beings arrived in the the river fed a large, permanent lake in the New World at the close of the Pleistocene, Manix Basin for much of the late Pleistocene no longer than 12,000 years ago, and that (Jefferson 1985). The lake attained a maxi­ Clovis sites represent the oldest occupation mum level of approximately 1,780 ft. (543 in the Americas (Haynes 1970; Martin 1973; m.), and drained before ca. 17,000 B.P. (Dorn Waters 1985). On the other hand, a less et al. 1986; Meek 1988). During extensive widely accepted school of thought sees a surface surveys around the ancient shorelines variety of evidence for human occupation in of this basin, particularly the 1,780-foot the Americas well back into the Pleistocene, shoreline, Simpson (1958, 1960, 1964) found a with dates ranging from 19,000 B.P. at Mea- variety of artifacts which she grouped to­ dowcroft Rockshelter, Pennsylvania (Stucken- gether as the Manix Lake Industry. She rath et al 1984), to 32,000 B.P. at Boquiero assigned these artifacts a late Pleistocene do Sitio da Pedra in Brazil (Guidon and age coincident with the maximum lake stand, Delibrias 1986), and to at least 220,000 B.P. thought at the time of her research to date at Calico Hills in the California desert between 12,000 and 25,000 B.P. (Bischoffetal. 1981). The Manix Lake Industry is characterized Unfortunately, much of the debate by large, roughly worked bifaces, as well as between these two positions is characterized by a few simple types of unifacially re­ by preliminary research reports, hasty touched flakes, hammerstones, "Clactonian" examinations of sites, a near absence of flakes, and discs. Specimens identified as published data, and unsystematic and incom­ Manix Lake implements principally are found plete analyses. This is particularly true in the eastern Calico Mountains adjacent to when the debate is over claims for early the Coyote Lake embayment of the Manix occupation in southern California, a region Basin, on desert pavement surfaces composed that has produced more purportedly early of volcanics and raw nodules of chert, chal­ material than any other part of North cedony, and jasper. Although Simpson noted America (cf. Moratto 1984). The purpose of that debris from lithic procurement and ini­ this paper is to present recent data bearing tial tool production occurred on these sites, on one of these claims, the Manix Lake she explicitly argued that this debris is Industry (Simpson 1958, 1960, 1964). distinct from the material she identified as

[209] 210 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY

>> X! UIM. J uss e O o 1—.. M -.-> x: .^ T) D-, QB tn O M &£ tn OJ uD .^ (0 ^H OQ ^^ o MANIX LAKE INDUSTRY 211 part of the Manix Lake Industry. lected only the most finished artifacts, Simpson's discussion of the Manix Lake creating a biased view of the material on Industry can be divided into two hypotheti­ the sites surveyed; (3) artifacts typologically cal arguments. The first is the argument identical to those found as isolates and that Manix Lake implements represent the designated ancient are commonly associated level of technology extant in North America directly with their primary reduction debris; during the period when they were made, that (4) embeddedness in a desert pavement is no is, that these implements were deposited guarantee of great antiquity; and (5) the around Manix Lake as finished pieces. The Manix Lake artifacts are not associated with second is the argument that these imple­ the high stand of the lake but with the ments represent a late Pleistocene occupa­ desert pavements containing the stone from tion. which they were made. The first position is supported principally Glennan supported these arguments by an by the following: (1) the occurrence of the analysis of new collections from the eastern artifacts as isolates rather than in direct edge of the Calico Mountains. Simpson association with the abundant manufacturing (1976) found the argument unconvincing. debris found elsewhere on the sites with them; (2) the degree of refinement of the DATING MANIX LAKE IMPLEMENTS artifacts identified as Manix Lake imple­ One of the major problems with assigning ments; and (3) the existence of edge damage a Pleistocene age to the Manix Lake Indus­ interpreted as evidence of use on some of try is the complete absence of direct chro­ the artifacts. nological information on Manix Lake imple­ The second position is supported by the ments. As the preceding section discusses, following: (1) the fact that the implements analyses to date have relied entirely on are embedded in desert pavements; (2) the inferential links between surface artifacts typological crudeness of the implements and whose age is unknown and geomorphic fea­ their formal similarity to Old World artifacts tures whose age is known. Recent work known to be of Pleistocene age; (3) the lack using cation-ratio dating, however, provides of projectile points on the sites that produce direct chronometric data on Manix Lake im­ Manix Lake implements; and (4) the apparent plements. Although a full report on this association of Manix Lake implements with work is forthcoming (Whitley and Dorn, the high stand of the lake, which is known unpublished data), its chronological results to date from the late Pleistocene. are directly relevant here. Like other purported pre-Clovis occupa­ Cation-ratio dating is a new method of tions in the New World, the Manix Lake dating surface artifacts recovered from arid Industry has not been accepted by most regions (Dorn 1983; Dorn and Whitley 1984; archaeologists. Glennan (1976; also see Dorn et al. 1986; Harrington and Whitney Wallace 1962) provided the most explicit 1987). This method is based on changes objections to both the dating and the inter­ over time in the chemical constituents of pretation of the Manix Lake material, rock varnish, an accretion on rock surfaces arguing that (1) the Manix Lake assemblages that is composed of clay minerals, oxides of are typologically crude because they are manganese and iron, and minor and trace el­ quarry and workshop debris rather than ements including organic carbon. The minor formed artifacts; (2) the collecting proce­ elements that accumulate in this varnish dures used in the Manix Lake surveys se­ have different susceptibilities to leaching by 212 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY meteoric water. Over time, moisture leaches the Mojave Desert (Dorn and Whitley 1984). away certain of these elements (cations) at a Six radiocarbon analyses were obtained on faster rate than others. the organic components of varnish after the Thus, there is a progressive change in curve was constructed, and the additional the ratio of the more mobile cations (e.g., ^^C/cation-ratio calibration points corrobor­ potassium [K] and calcium [Ca]) to more ated the initial results. stable cations (e.g., titanium [Ti]) over time. The curve used to derive the dates em­ By itself, a cation ratio (such as [K+Ca]/ ployed in later sections of this paper differs Ti) provides relative dates for varnished sur­ sli^tly from the one depicted in Figure 2, faces: higher ratios indicate younger ages. and these dates therefore vary slightly from However, the ratios can also be calibrated those reported elsewhere (Bamforth et al. by an absolute chronology. By measuring 1986; Dorn et al. 1986). This is because the cation ratio in varnish on geomorphic recent tests at the University of Arizona surfaces of known age, the age of the sur­ tandem accelerator (Ronald Dotn, unpub­ faces and the cation ratios can then be lished data on file at Arizona State Univer­ plotted against one another to construct a sity; Dorn et al. 1987) indicate that ^^C least-squares regression, called a cation- dates derived from desert varnish average leaching curve, which describes the rate at approximately 10% younger than the age of which unstable cations are leached over the underlying varnish surface. This time. The age of varnish whose antiquity is discrepancy apparently results from (1) a unknown can then be determined by plotting slight delay between the exposure of a its cation ratio on this curve, thereby pro­ surface to the varnishing process and the viding a minimum age for the underlying onset of that process; and (2) the fact that surface. Changes in cation ratios can also even using only organic material from the be calibrated by extracting the organic con­ lowest levels of the varnish for Tandem tent of the lowest (i.e., oldest) levels of the Accelerator Mass Spectrometry (TAMS) varnish on a surface and obtaining both a analysis includes material deposited over a radiocarbon date on this material and a period of time after the varnish began to cation-ratio reading. Such dates can also be form. To correct for this, the ages derived plotted against cation ratios to construct a from the TAMS analyses were increased by curve, and are especially useful where no 10% and a new calibration curve was derived. previous dates on geomorphic surfaces are The equation for this curve is Y= 12.21-1.90X, available. Unfortunately, however, there is where Y equals the cation ratio and X equals not enough carbon in varnish on artifacts the logio of the absolute age. Although for even an accelerator radiocarbon date some past applications of this technique have (Dorn et al. 1986). derived dates with fairly large error margins The cation-leaching curve for the central (i.e., Dorn and Whitley 1983, 1984), the dates Mojave Desert (Fig. 2) uses both of these for the core/flake sequences discussed here methods and was constructed using cation have a mean error margin of 11.0% (median ratios from youthful potassium-argon-dated = 8.0%). basalt flows in the Cima volcanic field, the Although the details of the method (Dorn radiocarbon-dated high stand of Silver Lake 1983; Dorn et al. 1986) and its application (Warren and Ore 1971), and modem aeolian here are discussed extensively elsewhere fallout, assigned an age of 100 B.P. based on (Bamforth et al. 1986), it is useful to estimates of the rate of varnish initiation in summarize the reasons for accepting the MANIX LAKE INDUSTRY 213 Years Before Present

10 10 10 10 10 rn~ ' 1—I—I I I I It I I Nil 1—I 1 I I I I I I

I I I I I I -I I ' • I I •' I I I I ' I ll -I I I I I i I Fig. 2. Initial cation-leaching curve for the east-central Mojave Desert, California. Numbers refer to K-Ar dates and letters refer to TAMS radiocarbon dates listed in Dorn et al. (1987:Table 1). The horizontal bars represent, respectively, two standard deviations of the age-uncer­ tainties and of the mean varnish cation-ratios for each calibration point. The error of the initial varnish ratio is represented as one standard error and is indicated by brackets on the left margin of the graph (see Dorn et al. 1987). The lines represent the first estimates of the semilog least-squares regressions indicating the probable rate of cation leaching in the east-central Mojave Desert (see text for a more recent estimate). 214 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY present calibration of the leaching curve fication, such as that by Harrington and constructed for the Mojave Basin and hence Whitney (1987) in New Mexico and Glazov- for arguing that the dates derived for this skiy (1985) in the Soviet Union. The cation- study are accurate. These reasons are as ratio dates derived for the IPP project dis­ follows: cussed below are therefore accepted for the 1. The archaeologically relevant portions remainder of this discussion with the caution of the curve are calibrated almost entirely that additional research may uncover prob­ by radiocarbon dates. lems not apparent. 2. The calibration points show an almost A total of 20 surface artifacts identified perfect correlation (r = -.99) between dates as Manix Lake bifaces by F. Budinger were and cation-ratios, documenting the linear na­ collected from alluvial fans in the eastern ture of the cation-exchange process for the Calico Mountains by R. Dorn, D. Whitley, time period relevant to archaeologists. Fur­ and F. Budinger and submitted for cation- thermore, three new radiocarbon dates for ratio dating, along with one other biface the central Mojave region provide additional identified by R. Simpson. These artifacts support for the calibration (Ronald Dorn, comprise a select, nonrandom sample of ma­ unpublished data on file at Arizona State terial intended to represent a range of University). macroscopic varnish development, from al­ 3. There is a change in varnish structure most complete surface coverage to barely at the Holocene/Pleistocene boundary in the visible. Mojave Desert (Bamforth et al. 1986). This The dates derived from these 21 artifacts change is visible in artifacts cation-ratio range from less than 400 to 32,000 B.P., with dated to the later Pleistocene but not in 12 of them producing dates within the last artifacts dated to the Holocene. 10,000 years and nine producing older dates. 4. Radiocarbon and cation-ratio dates on These results suggest that human occupation varnish from a late-Pleistocene (varnish ra­ of the region may in fact have occurred in diocarbon age of ca. 14,600 B.P.) basalt flow late Pleistocene (cf. Dorn et al. 1986), a in the Cima volcanic field correspond with point to which we return below, but they age estimates based on soils data (McFadden show fairly clearly that the material pres­ et al. 1986) and paleomagnetic fluctuations ently grouped together as the Manix Lake (D. Champion, personal communication Industry does not represent a single period 1986). of human occupation in the region and 5. The cation-ratio dates for individual therefore that the criteria used to group this components of refitted cores are all consis­ material are not temporally meaningful. In tent with one another; that is, no conjoin- the next two sections of this paper, we pre­ able artifacts produced noticeably different sent additional data which both support this ages. conclusion and also account for those as­ There are simply no data currently avail­ pects of the distribution of "Manix Lake" able to contradict this evidence, although material that have been argued to indicate Dorn et al. (1986) noted several theoretically their great age. possible complications with the method. The major reason for treating the dates pre­ THE IPP PROJECT sented here as tentative is the experimental One of the major problems in evaluating nature of the method, a problem that can be the various interpretations of the Manix solved only by additional independent veri­ Lake Industry is the general absence of MANIX LAKE INDUSTRY 215 detailed supporting data. Simpson's various these areas were not used due to possible discussions offered summaries with little bias introduced by the use of different field supporting evidence, while Glennan's (1976) methods. The major foci of our attention discussion included information on only a were the generally well-defined clusters of small portion of the area of interest. How­ flakes, cores, and bifaces that dominate the ever, a substantial body of data directly material found on the IPP sites. Although relevant to the Manix Lake debate was ob­ isolates were recorded, collections within the tained as part of the work performed to right-of-way were essentially confined to mitigate impacts of the California segment of these clusters. the Intermountain Power Project (IPP). Four Traditional methods of dating are not quarry sites located on desert pavement useful for the IPP material, which consists quarries between Baker and Barstow were in­ almost entirely of surface-collected flakes vestigated, including two overlooking the and cores. Nor is this material in itself Manix Basin. One of these, CA-SBr-2100, is temporally diagnostic. We therefore relied immediately adjacent to the Calico Moun­ on cation-ratio dating (see above) to obtain tains and is one of the localities where the absolute dates on individual artifacts in the Manix Lake Industry was first recognized. collection. Only specimens from refitted Bamforth et al. (1986) and Bamforth (n.d.) cores were dated in order to eliminate the presented a complete discussion of the possibility of obtaining extremely old dates results of this project and a reconstruction on naturally produced flakes. All of the of prehistoric human activities on the artifact clusters collected during the project project sites; our concern here is with only were refitted insofar as possible, and dates those aspects of the IPP data relevant to were obtained for all refits whose constitu­ the Manix Lake Industry debate. ents bore adequate amounts of varnish; a Three aspects of this project are relevant total of 167 artifacts from 66 refits were here. The first two are a mapping program dated. These dates were summarized in covering the full extent of each of the sites Dorn et al. (1986) and were discussed in investigated and the subsequent collection of detail in Bamforth et al. (1986). artifacts within the IPP right-of-way. The THE IPP DATA third is the information derived from cation- ratio dating of portions of this collection. As was discussed above, the identification The mapping work was based on a two- of the Manix Lake Industry as evidence of a dimensional systematic, unaligned sampling Pleistocene occupation in the Mojave Desert procedure (e.g., Cochran 1977:277-278) in rests on two major supporting ideas: the which one unit 40 m. in diameter was placed typological similarities of the Manix Lake within each 200-yd. block in a grid laid artifacts to Pleistocene artifacts from the down over each site, providing a 3.76% sam­ Old World and the occurrence of those arti­ ple of the area investigated. All cultural facts as isolates embedded in desert pave­ material within each sample unit was ments around the high stand of Pleistocene mapped. This sampling procedure was used Lake Manix. We have presented data above on three of the four quarries studied for the which indicate that technologically crude project: CA-SBr-2100, CA-SBr-2223, and bifaces are not temporally diagnostic in the half of CA-SBr-3183. The fourth site (CA- Manix Lake region, a point which is true for SBr-2162) and the other half of CA-SBr-3183 all of North America (cf. Stanford 1982:205). were surveyed completely, but the data from The material collected and dated for the 216 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY

IPP project, mainly cores and unmodified surface over time, we should tend to see flakes, generally is not typologically part of older dates associated with clusters more the Manix Lake Industry, although many of deeply buried. Class 1 clusters, with the the artifacts located during the site-mapping highest proportion of surface artifacts, phase of the project might be classified as should therefore tend to be youngest, and belonging to it. The remainder of this sec­ Class 4 clusters, with the highest proportion tion therefore examines the second support­ of deeply buried artifacts, should be oldest. ing idea just noted, the embeddedness and Table 1 presents the mean ages of dated isolation of the Manix Lake implements and artifacts from clusters in each of these four their association with the shoreline of the classes, along with the maximum and mini­ maximum stand. mum ages obtained for any cluster. (These Embeddedness in Desert Pavement maximum and minimum figures refer to the oldest and youngest ages obtained from in­ The material collected during the IPP dividual refitted cores within each class, and project was embedded to various depths in not to the errors associated with the dates the desert pavement. The depth of embed­ on any single refit.) It is obvious that ding (or artifact burial) was recorded for the there is no particular relationship between artifacts in most of the clusters collected. the four classes of clusters and the mean This information unfortunately cannot be age of refitted artifacts from them. linked to individual artifacts because of time Other aspects of the data suggest that constraints in the field, but it can be evalu­ the major factors controlling the degree of ated on a cluster-by-cluster basis. Artifact burial of artifacts in a cluster are the burial data are available for 36 of 49 clus­ degree of slope and nature of the surface on ters that produced dated, refitted cores. which the artifacts were deposited. Table 2 Degree of burial was assessed in five cate­ shows the frequencies of the four classes of gories: surface, 1% to 25% of the artifact clusters for these two variables. Slope was buried, 26% to 50% buried, 51% to 75% evaluated in the field and dichotomized into buried, and greater than 75% buried. level/gentle and moderate/steep. Surface Visual inspection of the frequency distri­ type was also evaluated in the field and was butions of the degree of burial of the flakes divided into armored pavement, semi- within the 36 clusters showed that they fall armored pavement, and nonpavement. The generally into four groups, or classes. Repre­ sample of clusters is too small in many of sentative examples of these four classes are the categories for detailed analysis, but it is presented in Figure 3. Class 1 is comprised clear that the overall tendency is for more of those clusters with most of the artifacts deeply buried clusters to be found on steeper on the surface and rapidly diminishing fre­ slopes and more loosely compacted surfaces. quencies of artifacts more deeply buried. Slopewash (Sharon 1962), pavement redistri­ Clusters with most artifacts buried from 1% bution processes caused by upthrust (Sprin­ to 25% are in Class 2. Class 3 includes ger 1958), and the accumulation of desert clusters with artifact frequencies increasing loess (Mabbutt 1979; McFadden et al. 1986) from the surface to greater than 25 percent probably are responsible for this pattern. burial. Class 4 is a miscellaneous group Controlling for local depositional condi­ whose members all have a high proportion of tions should show chronological relationships artifacts buried more than 50 percent. If if they exist. Within each combination of artifacts tend to become embedded in the conditions, Simpson's hypothesis implies that MANIX LAKE INDUSTRY 217

Cluster 118, SBr 2100 Cluster 37, SBr 3183

20- 20- CLASS 1

(N) 10 (N) 10-

0 25 50 T=¥75 100 % 0 26 50 75 100% (N) = NUMBER OF ARTIFACTS

Cluster 118, SBr 2100 Cluster 194, SBr 2223

20- 20- CLASS 2 (N) 10- (N)(0- -

^F=f^ 1 0 26 60 76 100% 0 26 60 75 100%

Cluster 85, SBr 3183 Cluster 54, SBr 3183

20- 20- CLASS 3

(N)IO- (N) 10- • =U 0 25 50 75 100% 0 25 so 75 100%

Cluster 49. SBr 3183 Cluster 148, SBr 2223

20- 20- CLASS 4

(N) 10 (N)IO-

0 25 50 75 100% 0 25 50 75 100%

Fig. 3. Examples of artifact clusters on the IPP sites classed by overall degree of embeddedness in the desert pavement, where N equals the number of artifacts. 218 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY

Table 1 MEAN AGES IN YEARS B.P. OF ARTIFACTS BY DEPTH OF EMBEDDING

Depth Class' N Mean Age Maxiinum Age Minimum Ag< 1 10 4,321 11,815 492 2 17 1,612 8,756 298 3 8 5,592 21,643 375 4 15 3,964 18,445 565

Class 1: most artifacts on the surface; Class 2: 1% to 25% burial of most artifacts; Class 3: 25% to 50% burial of most artifacts; Class 4: 50% or greater burial of most artifacts.

Table 2 FREQUENCY OF COLLECTED CLUSTERS BY DEPTH OF EMBEDDING AND LOCAL DEPOSFHONAL CONDTOONS

Depth Local Depositional Conditions Class' Armored Semi-armored Flat Semi-armored Slope Nonpavement Total 1 3 6 2 10 14 3 3 8 1 8

Total 17 36

Class 1: most artifacts on the surface; Class 2: 1% to 25% burial of most artifacts; Class 3: 25% to 50% burial of most artifacts; Class 4: 50% or greater burial of most artifacts.

Table 3 MEAN AGES IN YEARS B.P. OF DATED ARTIFACTS BY DEPTH OF EMBEDDING AND LOCAL DEPOSmONAL CONDinONS Depth Local Depositional Conditions Class* Armored Semi-armored Flat Semi-armored Slope Nonpavement 1 6,259 3,686 2,810 2 1,625 1,278 3,985 3 8,298 585 8,530 4 1,694 3,989 3,645

Class 1: most artifacts on the surfiace; Class 2: 1% to 25% burial of most artifacts; Qass 3: 25% to 50% burial of most artifacts; Class 4; 50% or greater burial of most artifacts. the four classes of clusters should show came because several clusters have produced increasing age. As shown in Table 2, there as many as four dated refits separated by is no apparent relationship between depth several thousand years. However, if there is class (or degree of embeddness) and deposi­ a strong relationship between age and arti­ tional conditions. The cation-ratio dates fact burial, at least a general association cannot be taken as direct estimates of the between the classes of clusters defined here age of the entire cluster from which they and increasing age should be apparent. No MANIX LAKE INDUSTRY 219 such association is evident in Table 3. results of Bowers et al. (1983) indicate Furthermore, even the Class 4 clusters clearly that clusters of this size and density (those with the highest proportion of deeply could readily disperse to the point where buried artifacts) are commonly associated they would be unrecognizable within a few with Holocene dates. thousand years, even given a very conserva­ The IPP data thus indicate that local tive rate of artifact movement. depositional conditions and not age are the Larger objects, such as cores and bifaces, major factors conditioning the degree to should move more readily on steeper slopes which artifacts are incorporated into the than on more gentle ones, while smaller ob­ desert pavements at the sites investigated: jects such as unmodified flakes should move even relatively deeply embedded artifacts can relatively easily on all slopes (cf. Carson and be of Holocene age. Embeddedness in a Kirkby 1972). If the isolates on the IPP desert pavement therefore does not provide a sites are the remnants of dispersed clusters, reliable basis for drawing chronological then intact clusters, isolated cores and bi­ inferences (cf. Glennan 1976). faces, and isolated flakes should show dis­ tinct distributions. Clusters should show the Distribution of Isolates most restricted distribution, being largely Simpson (1960) repeatedly has asserted confined to the intact pavements towards the that some of the best evidence for the hy­ centers of the sites. Isolated cores and pothesis that the Manix Lake artifacts indi­ bifaces should be more dispersed across the cate habitation areas rather than quarry site areas, and isolated flakes should be the activity is that they are not associated with most dispersed of aU. This prediction was nearby workshops. However, none of the tested for the three IPP sites on which participants in the Manix Lake debate has sample units were placed: CA-SBr-2100, CA- ever reported the full distribution of iso­ SBr-2223, and CA-SBr-3183. The last of lates. During the IPP project, many isolates these three sites is on flatter terrain than were recorded and mapped. These data sug­ the other two, and the distributions of gest that the significance of the distribution clusters, cores and bifaces, and flakes on it of isolates on desert pavement sites cannot should be more alike than on the other two, be evaluated unless post-depositional pro­ which show a considerable degree of topo­ cesses are taken into account. Given that graphic relief. discrete clusters of primary production debris The data support these predictions. The are the dominant type of remains found on degree of artifact and cluster dispersion on such sites, it is particularly critical to CA-SBr-2100, CA-SBr-2223, and CA-SBr- consider what happens to such clusters when 3183 can be assessed quantitatively, but the they are exposed on the surface for extend­ nature of the available data affects the ed periods of time. methods that can be used for such an as­ Experimental research by Bowers et al. sessment. A simple way of measuring the (1983) indicates that surface clusters of degree of dispersion of a set of entities flakes tend to disperse over time as the (e.g., clusters, cores, bifaces, or flakes) cumulative result of small, random move­ across an area that has been sampled is to ments of the individual artifacts they con­ compute the ratio of the variance of the tain. Clusters on the IPP sites typically number of observations of like items per contain from 10 to 200 artifacts dispersed sample unit to the mean number of observa­ over areas of 5-15 m.^ The experimental tions per sample unit, thus deriving a statis- 220 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY tic known as the Coefficient of Dispersion more, on CA-SBr-2100 and CA-SBr-2223, iso­ (Greig-Smith 1983:61-62; see Thomas [1975] lated flakes are dispersed noticeably more for an archaeological application). The than either of the other classes of material. variance of a uniformly dispersed set of On CA-SBr-3183 all three classes of material observations will be zero, while the variance show essentially identical degrees of dis­ of a clumped distribution will be relatively persion, indicating that the substantially high, and the variance/mean ratio will lower and more continuous slope in the sam­ therefore be lower for uniformly dispersed pled area of this last site has reduced the than for clumped distributions. degree of artifact movement. However, analyses based on the variance These patterns thus conform to the ex­ of a nonnormal distribution can be seriously pectations discussed above for the effects of misleading, and this is a particular problem post-depositional dispersion of previously for the data at issue here. The distributions discrete artifact clusters over time. The of observations of aU classes of material per occurrence of an artifact as an isolate may sample unit are extremely right-skewed (Fig. or may not indicate that it is older than an 4), and such a degree of skew artificially artifact that occurs in a nearby, well- inflates the variance. Furthermore, because defined cluster, but the data presented here the values of observations within classes strongly suggest that it cannot be taken as vary considerably, the degree to which the evidence that such an artifact is of Pleisto­ variance is inflated differs between classes. cene age. The likelihood that isolates on Comparisons between classes of material the IPP sites were once part of clusters that using the variance are therefore meaningless have dispersed over time fits well with in this case. Glennan's (1976) observation that artifacts However, the essential meaning of the typologically identical to isolates also occur Coefficient of Dispersion, although not its as parts of clusters. strict probabilistic interpretation, can be captured for comparative purposes by using Association with the Shoreline statistics which are more resistant to distor­ The third major point of contention in tion. For this analysis, we can substitute the Manix Lake Industry debate discussed the interquartile distance (Hartwig and here is whether the artifacts included in the Dearing 1979:21) for the variance and the industry are associated with the 1,780-ft. median for the mean and take the ratio of shoreline of Manix Lake or with the desert these nonparametric alternatives. Again, this pavement that produced the material from adaptation of the Coefficient of Dispersion is which they are made and that is only coin- simply a convenient device for comparing the cidentally above that shoreline. The latter relative degree of dispersion of the different position, taken by Glennan (1976), clearly classes of material on the three sites. implies that the density of artifacts should Table 4 presents the results of this be much higher on the pavements than off analysis, which clearly conform to the pat­ them, with proximity to the 1,780-ft. shore­ terns predicted above. First, the degree of line having no effect on this relationship. dispersion of clusters is similar to that of Conversely, if artifacts are associated with cores and bifaces on all three sites, and on the shoreline, they should be evenly dis­ the only site where a difference between the tributed on and off the pavements above the two is visible (CA-SBr-2100), the latter are 1,780-ft. level If isolates are the dispersed more dispersed than the former. Further­ remnants of clusters, and the clusters are MANIX LAKE INDUSTRY 221

CD

Clusters/Sample Unit

c 3

Q. E o

0) E

10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 Isolated Cores end Bifoces/Scmple Unit

c 3

a. E o 00

0) X) E

Isolated Fiakes/Sannple Unit

Fig. 4. Frequenqr distributions of numbers of intact clusters, isolated cores and bifaces, and isolated flakes per sample unit on CA-SBr-2100. 222 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY

Table 4 DEGREE OF DISPERSION OF CLUSTERS AND ISOLATED ARTIFACTS ON CA-SBR-2100, CA-SBR-2223, AND CA-SBR-3183 Interquartile Site Class Median Distance Dispers CA-SBr-2100 Intact clusters 3.0 7.0 2.3 Isolated cores and bifaces 6.0 11.0 1.8 Isolated flakes 37.0 45.0 1.2

CA-SBr-2223 Intact clusters 3.0 6.0 2.0 Isolated cores and bifaces 2.0 4.0 2.0 Isolated flakes 24.0 36.0 1.1

CA-SBr-3183 Intact clusters 2.0 4.0 2.0 Isolated cores and bifaces 2.0 4.0 2.0 Isolated flakes 10.5 22.0 2.1

interquartile distance/median

Table 5 COMPARISON OF FREQUENCY OF CLUSTERS AND ARTIFACTS FROM ON AND OFF PAVEMENT AT THREE SITES On Pavement Off Pavement Site N» Cluster Mean± SD Artifact Mean± SD N« Cluster Mean ± SD Artifact Mean ± SD CA-SBr-2100 28 8.4±5.7 81.7±64.6 17 0.2 + 0.5 21.4 ±13.8 CA-SBr-2223 24 5.9 + 3.9 43.1 + 39.5 22 0.3 ±0.7 21.4 ±24.9 CA-SBr-3183 8 4.0 ±2.6 24.8 ±18.3 9 0.0 ±0.0 3.0 ±4.9

N equals the number of sample units. directly associated with the pavements, the only those for which this information was relationship between pavements and cluster clearly indicated. The pattern shown by densities should be somewhat stronger than these data is obvious: both clusters and for the isolate densities, as isolates may isolated artifacts are overwhelmingly more have washed off the pavements onto the abundant on the pavements than off. This surrounding flats. pattern is strongest for clusters, which are All of the sample units on CA-SBr-2100 42 times more abundant on than off pave­ and CA-SBr-2223 were placed above the ments on CA-SBr-2100, 19 times more fre­ 1,780-ft. level; this level is not relevant to quent on pavements on CA-SBr-2223, and are CA-SBr-3183 as the latter site is not in the associated only with pavements on CA-SBr- Manix Basin. CA-SBr-2162 is not included 3183. Although the difference in the fre­ here because it was not surveyed with sam­ quencies of isolates on and off pavements is ple units. Table 5 compares mean cluster not as pronounced, as we predicted above, it and artifact frequencies per sample unit for is clear nonetheless. units on and off the pavements. Unfortun­ SUMMARY AND CONCLUSIONS ately, whether or not a unit was on a pavement was not recorded in every in­ Two basic hypotheses can be defined in stance; the units included in Table 5 are the debate over the Manix Lake Industry. MANIX LAKE INDUSTRY 223

The first, championed by Simpson (1958, parison provides additional support for these 1960, 1964), holds that certain isolated interpretations. artifacts found embedded in desert pave­ The data presented here thus indicate ments above the 1,780-ft. shoreline of that the Manix Lake Industry, as heretofore Pleistocene Manix Lake represent a late characterized, cannot be taken as evidence Pleistocene lakeside human occupation. The for Pleistocene human occupation in the New second, most explicitly framed by Glennan World. The cation-ratio dates on supposed (1976), holds that these artifacts are quarry Manix Lake implements show no tendency to debris found on desert pavements fortu­ cluster in the Late Pleistocene. Critical itously above the 1,780 ft. level, and that points in support of the Manix Lake hy­ embeddedness in a pavement is no guarantee pothesis have included the embeddedness of of great antiquity. Manix Lake implements in desert pavements, Cation-ratio dating of supposed Manix their occurrence as isolates, and their ap­ Lake implements shows no evidence that the parent association with the high stand of criteria used to identify such implements are Manix Lake. These can all be accounted for temporally significant: cation-ratio dates on as either the result of post-depositional Manix Lake material range from less than processes or the likelihood that the Manix 400 to 32,000 years B.P. In addition, other Lake material is quarry debris associated data on the distribution of artifacts near with the pavements on which raw lithic ma­ Manix Lake and elsewhere indicate that the terial occurs. patterns argued to support a great age for Ironically, the studies reported here have Manix Lake implements are the result of seemingly produced evidence of a late Pleis­ post-depositional processes. First, these tocene human presence in the Mojave Desert. data show no relationship between age and We have noted cation-ratio dates as old as the overall degree of embeddedness of a 32,000 B.P. on rough bifaces from the east­ cluster in the pavement. Rather, the overall ern Calico Mountains, and the IPP project degree of embeddedness of the clusters produced four refitted cores dated by the seems to be a function of the degree of cation-ratio method from 14,000 to 22,000 slope and the development of the pavement B.P. (see Bamforth et al. 1986:101-105; Dorn on the surface of which they were found. et al. 1986). However, "Manix Lake" Second, the distribution of artifacts con­ implements (bifaces apparently dating from forms to that expected if isolates are largely the Pleistocene) and bifaces dated through­ the products of post-depositional dispersion out the Holocene were identified using the of clusters of quarry debris. Third, there is same criteria. The IPP material differs in an unambiguous association of isolates with no way from the Holocene material in the desert pavements regardless of proximity to project sites: like the rest of the IPP the 1,780-ft. shoreline. These patterns are collection, it clearly is early-stage quarry found on CA-SBr-2100, purported to be a reduction debris. We provisionally accept major Manix Lake Industry site, CA-SBr- these dates because, as we discussed earlier, 2223, which also overlooks the Manix Basin, there are no good reasons for rejecting any and CA-SBr-3183, which is not directly of the Mojave River Valley cation-ratio associated with any Pleistocene lake. The dates. We believe that the irony of this existence of essentially identical patterns on situation simply emphasizes the distinction both a supposed Manix Lake Industry site between chronological inferences drawn from and on the two sites included here for com­ direct chronometric information and chrono- 224 JOURNAL OF CALIFORNLA. AND GREAT BASIN ANTHROPOLOGY logical inferences drawn without the benefit Carson, M., and M. Kirkby of such information. We hope that our 1972 Hillslope Form and Process. Cambridge analysis has demonstrated how cautious Geographical Studies No. 3. archaeologists must be when working in the Cochran, W. G. latter situation. 1977 Sampling Techniques (3rd edition). New York: John Wiley and Sons. ACKNOWLEDGEMENTS Dorn, R. 1983 Cation-Ratio Dating: A New Age-Deter­ Archaeological research for the Intermountain mination Technique. Quaternary Re­ Power Project was funded by the Los Angeles search 20:49-73. Department of Water and Power and benefited from the archaeological insight and management Dorn, R., D. Bamforth, T. Cahill, J. Dohrenwend, abilities of Mike Macko, Ed Weil, and Gary B. Turrin, D. Donahue, A. Jull, A. Long, M. Dudley of Applied Conservation Technology. Macko, E. Weil, D. Whitley, and T. Zabel Peter Ainsworth, Neil Rhodes, and Bob Nagle 1986 Cation-Ratio and Accelerator Radiocar­ refitted quarry debris for many, many hours; bon Dating of Rock Varnish on Mojave much of this analysis was possible only because Artifacts and Landforms. Science 231: of their abilities. The cation-ratio leaching 830-833. curve used to date the material discussed here Dorn, R., and T. M. Oberlander integrates calibration points derived from the 1982 Rock Varnish. Progress in Physical IPP research and from other research conducted in the project area, including some completed Geography 6:317-367. under NSF grant SES-86-01937. Fred Budinger Dorn, R., D. Tanner, B. Turrin, and J. Dohren­ of the Calico Early Man Project and Dave wend Whitley of the University of Witwatersrand 1987 Cation-Ratio Dating of Quaternary Ma­ assisted in the collection of the Manix Lake terials in the East-Central Mojave bifaces reported in preliminary form here. Desert, California. Physical Geography 8:72-81. REFERENCES Bamforth, D. B. Dorn, R., and D. Whitley n.d. Settlement, Raw Material, and Lithic 1983 Cation-Ratio Dating of Petroglyphs Procurement in the Central Mojave from the Western Great Basin, North Desert. Journal of Anthropological America. Nature 302:816-818. Archaeology (in press). 1984 Chronometric and Relative Age Deter­ Bamforth, D. B., R. I. Dorn, M. Macko, and E. B. mination of Petroglyphs in the Western WeU United States. Annals of the Associa­ 1986 Tradition and Transmission: Prehistoric tion of American Geographers 74:308- Stone Procurement and Tool Production 322. in the Mojave Desert, California. MS Glazovskiy, A. on file at the San Bernardino County 1985 Rock Varnish in the Glacierized Regions Archaeological Information Center, San of the Pamus. Data of the Glacierolog- Bernardino County Museum, Redlands. ical Studies 54:136-141. Bischoff, J., R. Shlemon, T. Ku, R. Simpson, R. Gleiman, W. Rosenbauer, and F. Budinger 1976 The Manix Lake Industry: Early Lithic 1981 Uranium-Series and Soil-Geomorphic Tradition or Workshop Refuse? Journal Dating of the Calico Archaeological of New World Archaeology l(7):43-62. Site, California. Geology 9:576-582. Greig-Smith, P. Bowers, P., R. Bonnichsen, and D. Hoch 1983 Quantitative Plant Ecology (3rd edition). 1983 Flake Dispersal Experiments: Noncul- Boston: Blackwell Scientific PubUca- tural Transformation of the Archaeolog­ tions. ical Record. American Antiquity 48: Guidon, N., and G. Delibrias 553-572. 1986 Carbon-14 Dates Pomt to Man m the MANIX LAKE INDUSTRY 225

Americas 32,000 Years Ago. Nature Sharon, D. 321:769-771. 1962 On the Nature of Hamadas in Israel. Harrington, C, and J. Whitney Zeitschrift fur Geomorphologie 6:129- 1987 Scaiming Electron Microscope Method 147. for Rock Varnish Dating. Geology 15: Simpson, R. 967-970. 1958 The Manix Lake Archaeological Survey. Hartwig, F., and B. E. Dearing The Masterkey 32:4-10. 1979 Exploratory Data Analysis. Beverly 1960 Archaeological Survey of the Eastern Hills: Sage Publications. Cahco Mountams. The Masterkey 34: Haynes, C. 25-35. 1970 Geochronology of Man-Mammoth Sites 1964 The Archaeological Survey of Pleisto­ and Their Bearing on the Origin of the cene Manix Lake (An Early Lithic Llano Complex. In: Pleistocene and Re­ Horizon). Proceedings of the 35th cent Environments of the Central Great International Congress of Americanists Plains, W. Dort and J. K. Jones, eds., 35:5-9. pp. 77-92. Lawrence: University of Kansas, Department of Geology Special 1976 A Commentary on William Glennan's Publication No. 3. Article. Journal of New World Archae­ ology l(7):63-66. Jefferson, G. 1985 Stratigraphic and Geologic History of Springer, M. the Pleistocene Manix Formation, Cen­ 1958 Desert Pavement and Vesicular Layer of tral Mojave Desert, California. In: Some Soils of the Lahontan Basin, Geologic Investigations Along Interstate Nevada. Proceedings of the Soil 15, Cajon Pass to Manix Lake, Califor­ Science Society of America 22:63-66. nia, R. Reynolds, ed., pp. 157-169. Redlands: San Bernardino County Stanford, D. Museum. 1982 A Critical Review of Archaeological Evidence Relating to the Antiquity of Human Occupation of the New World. Mabbut, J. A. In: Plains Indian Studies, D. H. 1979 Pavements and Patterned Ground in the Ubelaker and H. J. Viola, eds., pp. 202- Australian Stony Deserts. Stuttgarten 218. Smithsonian Contributions to Geographische Staten 93:107-123. Anthropology No. 30. Martin, P. 1973 The Discovery of America. Science 179: Stuckenrath, R., J. Adovasio, J. Donahue, and R. 969-974. Carlisle 1984 The Stratigraphy, Cultural Features, and McFadden, C, S. Wells, and J. Dohrenwend Chronology at Meadowcroft Rock Shel­ 1986 Influence of Quaternary Climatic ter, Washington County, Southwestern Changes on Processes of Soil Develop­ Peimsylvania. In: Meadowcroft: Col­ ment on Desert Loess Deposits of the lected Papers on the Archaeology of Cima Volcanic Field, California. Catena Meadowcroft Rock Shelter and the 13:361-389. Cross Creek Drainage, R. Carlisle and J. Adovasio, eds., pp. 69-90. Pittsburgh: Meek, N. University of Pittsburgh, Department of 1988 A Hypsometric Analysis of the Lake Anthropology. Manix Basin, California. Abstracts of the Phoenix meeting of the Association Thomas, D. H. of American Geographers (in press). 1975 Non-Site Sampling in Archaeology: Up the Creek Without a Site? In: Moratto, M. Sampling in Archaeology, J. Mueller, 1984 California Archaeology. Orlando: Aca­ ed., pp. 61-81. Tucson: University of demic Press. Arizona Press. 226 JOURNAL OF CALIFORNIA AND GREAT BASIN ANTHROPOLOGY

Wallace, W. J. Late Glacial North America, J. I. Mead 1%2 Prehistoric Cultural Development in the and D. J. Meltzer, eds., pp. 125-143. Southern California Deserts. American Orono: University of Maine Center for Antiquity 28:172-180. the Study of Early Man. Waters, M. R. Wilmsen, E. 1985 Early Man in the New World: An 1965 An Outline of Early Man Studies in the Evaluation of the Radiocarbon Dated United States. American Antiquity 31: Pre-Clovis Sites in the Americas. In: 172-192. Environment and Extinction: Man in