GEA(Wiley)

RIGHT

BATCH

Radiocarbon Dating of Bone Apatite by Step Heating Todd A. Surovell Department of Anthropology, University of Arizona, Tucson, Arizona 85721

Many advances have been made in radiocarbon dating of bone organics, but research on 14C dating of bone inorganic carbon has lagged significantly behind. Using mammoth bone, enamel, and tusk from the Dent and Murray Springs Clovis sites, experiments with the Haas and Banewicz technique of bone apatite dating by step heating demonstrate that accurate radiocarbon ages can be produced from bone apatite carbonate. Furthermore, preliminary findings suggest a correlation between the degree of apatite contamination and the slope of temperature-age spectra, providing a possible means of independently evaluating the accuracy of radiocarbon dates produced by this method. 䉷 2000 John Wiley & Sons, Inc.

INTRODUCTION Among his many accomplishments in the fields of archaeology and geology, C. Vance Haynes, Jr. has been involved in efforts to obtain reliable radiocarbon age determinations from bone and its various constituents from at least 1967 until the present. His early works on dating bone carbonates and organics are well known and frequently cited (Haynes, 1967, 1968), but his contributions have not ended there. Since that time, he has played the role of collaborator, advisor, or facilitator for a number of bone dating studies including this one (e.g., Haas and Banewicz, 1980; Hassan, 1976; Hassan et al., 1977; Pigati, 1996; Stafford et al., 1982, 1987, 1988). In this article I present a brief history of radiocarbon dating of bone apatite carbonate, followed by the results of preliminary experiments with the Haas and Banewicz technique of 14C dating of bone inorganic carbon by step heating. From the beginning, radiocarbon dates on bone mineral were destined to be fraught with difficulties. In 1955, even though Libby admitted to having no experience with bone apatite dating, he predicted that bone carbonate as a dateable material was “a very poor prospect for two reasons: the carbon content of bone is extremely low, being in largely inorganic form in a very porous structure; and it is extremely likely to have suffered alteration.” He continued his prophecy by writing, “It is barely conceivable that measurements on bone might reveal that some reliability could be obtained” (Libby, 1955:45). By the mid-1960s Libby’s prediction had been verified, and bone apatite dates were widely considered to be unreliable (Berger et al., 1964; Tamers and Pearson, 1965). In 1968, C. Vance Haynes, Jr. published a short article in Science that proved to be an important first step in improving the dating of bone mineral. He innovated the use of a preliminary acetic acid wash in pretreatment that takes advantage of Geoarchaeology: An International Journal, Vol. 15, No. 6, 591– 608 (2000) 䉷 2000 John Wiley & Sons, Inc.

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

the lower chemical stability of CaCO3 relative to bone apatite (Haynes, 1968). CaCO3 is a common contaminant in radiocarbon samples, and previous pretreatments, which generally employed HCl to hydrolyze bone samples, failed to distinguish between carbonates indigenous to the bone mineral and those secondarily deposited as CaCO3. With the addition of the acetic acid step, it was conclusively demonstrated that carbonate exchange could occur within bone apatite as Libby had suspected (Haynes, 1968). The next improvement in apatite dating came with a technique known as fractional hydrolysis, explored by Afifa Hassan in her doctoral dissertation (Hassan, 1976; Hassan et al., 1977). Fractional hydrolysis, like previous methods, uses strong acid to hydrolyze the bone mineral and emit carbonates as CO2. The CO2, however, is collected in successive fractions on the assumption that the acid first attacks the outer surfaces of bone crystals which should be more susceptible to contamination by exchange. The carbonates internal to the apatite crystals are released later and should be more pristine. While Hassan (1976) was able to improve 14C dates on bone apatite, those dates were still far from actual age of the sample. Attempting to improve upon Hassan’s technique, Haas and Banewicz (1980) argued that the failure of fractional hydrolysis to produce correct radiocarbon ages was due largely to difficulties relating to sample treatment. That is, hydrolysis was not attacking the outer surfaces of individual apatite crystals as intended, but instead was attacking the surfaces of the particles created by grinding samples in pretreatment. As an alternative, Haas and Banewicz advocated the use of step heating to release apatite carbonates because thermal energy acts largely independent of the geometric properties of the sample. Similar to fractional hydrolysis, this method relies on the assumption that contaminant carbonates will be released from the sample at lower temperatures, while the indigenous carbonates from unaltered apatite will be released at higher temperatures. Using this technique, they produced four radiocarbon dates from bison bone apatite from the Hudson – Meng archaeological site. The first three dates showed a trend of increased age with greater temperature of extraction, and their third date (9670 ⫾ 660 B.P.), collected at 820⬚C, was consistent with a charcoal date from the site (9820 ⫾ 160 B.P.). Interestingly, their final date (7240 ⫾ 620 B.P.), collected at 920⬚C was the youngest of the series. They believed this sample might have been subjected to a small amount of modern contamination at some stage during sample preparation. Following successes in the use of stable carbon isotopes from tooth enamel for paleodietary reconstruction (e.g., Lee-Thorpe and van der Merwe, 1991; Quade et al., 1992), two studies explored the potential of enamel carbonate for 14C dating (Hedges et al., 1995; Pigati, 1996). Tooth enamel is considerably less porous and considerably more crystalline than bone, both of which reduce the potential for contamination by exchange. Both studies were able to constrain errors in 14C measurements to less than 6% equivalent modern carbon contamination, but neither was able to produce the known ages of the samples used. With the exception of this recent work (see also Saliege et al., 1995), there has been a conspicuous silence for the last two decades in bone apatite dating. Around

592

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING

the same time that Haas and Banewicz published their study, 14C dating by accelerator mass spectrometry became a reality. This opened new avenues for dating of bone organics because milligram and submilligram sized samples became workable. Bone organics do not suffer from problems of chemical exchange that have plagued radiocarbon dates on bone inorganic carbon. Also, bone organics and organic contaminants are packaged as distinct compounds, such as proteins and amino, humic, and fulvic acids that are easily separated using chemical and chromatographic techniques. This is not true of dating carbonate in bone mineral where the contaminant and the material to be dated have an identical chemical structure. Although it has been conclusively demonstrated that reliable radiocarbon ages can be obtained from various constituents of bone organic matter, it has also been shown that this is only true for bone from geochemical contexts that favor organic preservation (Holliday et al., 1999; Stafford et al., 1987, 1991; Taylor, 1994). Where collagen is not well preserved, bone remains undateable. This fact alone emphasizes the need for more research focusing on producing reliable radiocarbon dates from bone apatite, a substance that is common in the archaeological and geologic records, and one that can persist for thousands of years across a broad range of conditions. Despite showing significant promise, the Haas and Banewicz study has not been replicated or explored further in the 20 years since it was published. Herbert Haas did experiment with the method to some extent, but was limited by the large amount of bone needed to produce enough CO2 for conventional 14C dating (Haas, personal communication, 1998). The research presented in this article is an initial attempt at AMS 14C dating of bone apatite carbonate by step heating to assess the viability of the method as a reliable geochronological technique. Three questions guided this work: (1) Can the step heating method produce correct radiocarbon ages from bones of known age? (2) What independent means can be used to assess the validity of the radiocarbon ages produced on samples of unknown age? (3) What is the effect of various pretreatments on temperature-age spectra? Question two is perhaps the most important as we currently have no means of ascertaining how a radiocarbon date on bone apatite might relate to its actual age, except perhaps the generalization that bone dates are often, but not always, too young (Tamers and Pearson, 1965; Taylor, 1987, 1994). A classic example illustrating that dates on bone carbonate can be falsely old is the caribou tibia flesher from Old Crow in the Yukon Territory. The original date produced on the total inorganic carbon fraction (27,000 ⫹ 3000/⫺ 2000 B.P.) far exceeded the subsequent date on bone collagen that demonstrated the tool was less than 1500 years old (Irving and Harington, 1973; Nelson et al., 1986). CONSIDERATIONS OF THE HAAS AND BANEWICZ TECHNIQUE As discussed above, step heating uses thermal energy to incrementally release carbonates from bone apatite as CO2. It rests on the assumption that there should be a direct positive correlation between thermal and chemical stability. That is,

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

593

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

some carbonates within the lattices of bone apatite crystals are more susceptible to chemical exchange than others. Carbonates near crystal boundaries should be more likely to exchange carbon with dissolved carbonates or with atmospheric CO2 than those in the interiors of bone crystals. Likewise, highly exchangeable carbonates closer to the surfaces of apatite crystals should require less thermal energy to be liberated as CO2 when heated. Therefore, as step-heating proceeds, the CO2 released at low temperatures should be dominated by carbon from highly exchangeable carbonates having high levels of contamination, and CO2 released at higher temperatures should be dominated by carbon from less exchangeable, or perhaps uncontaminated carbonates. A similar technique has been used in 40Ar/ 39Ar dating since 1966 to assess the effects of the loss of 40Ar by diffusion. In this field, step heating coupled with diffusion theory has resulted in a robust geochronological technique (McDougall and Harrison, 1999). Applying step heating to dating carbonates from bone mineral, however, may not be as straightforward as implied above for two reasons. First, carbonate ions exist in bone apatite crystals in three distinct positions (LeGeros et al., 1969; Poyart et al., 1975; Rey et al., 1989). Carbonates can be found substituted for PO43⫺ ions, and to a much lesser extent, for OH⫺ ions. Some carbonates are also known to be present adsorbed on crystals boundaries along hydration layers. The relative thermal stability of each of these carbonate fractions is poorly known and could complicate the interpretation of temperature-age spectra. Second, apatite recrystallizes commonly as buried bone undergoes fossilization (Hedges and Millard, 1995; Sillen, 1989; Weiner and Bar-Yosef, 1990). Likewise, heating of bone is also known to cause recrystallization (Person et al., 1996; Shipman and Schoeninger, 1984; Stiner et al., 1995). This reordering and growth of bone crystals during diagenesis and in the lab during step heating could have some effect on temperature-age spectra by shifting the positions of individual carbonate ions relative to apatite crystal surfaces, but again it is difficult to predict what affect this may have. Until further study directly addresses these issues, they are assumed to be insignificant and are not dealt with here. METHODS In an attempt to continually improve pretreatment and step heating regimes, experimental procedures evolved as each set of dates was produced. Because the methods used in subsequent trials were derived from the relative success and/or failure of previous trials, this research is presented in the chronological order in which it was performed. General procedures are outlined in this section, but procedural details specific to each trial are discussed in the Results section with reference to the reasoning behind their adoption. For simplification, a summary of methods, broken down by trial, is presented in Table I. Mammoth bone, tooth enamel, and tusk dentine from two Clovis archaeological

594

VOL 15, NO. 6

short standard

GEA(Wiley)

Pretreatment

N

Site (Series)

Material

Oven Heat

AA26641-7 AA27984-7 AA28261-4 AA31318-24 AA32634-7 AA33648-54

7 4 4 7 4 7

Dent (1) Dent (2) Dent (3) Dent (4) Dent (5) Murray Springs

Cortical bone Enamel Enamel Enamel Cortical bone Tusk dentine

⫺ ⫹ ⫹ ⫺ ⫺ ⫺

1 12

Step Heating

HAc

NaOCl⫹ Hac

NaOCl

N2 Tent

⫹ ⫺ ⫺ ⫹ ⫺ ⫹

⫹ ⫺ ⫺ ⫹ ⫺ ⫹

⫺ ⫺ ⫺ ⫺ ⫹ ⫺

⫺ ⫺ ⫺ ⫹ ⫺ ⫺

Abbreviations: HAc, acetic acid; N2 Tent, solution transfers performed under a nitrogen atmosphere.

Min. Temp. (⬚C)

Max. Temp. (⬚C)

Approx. Temp. Interval (⬚C)

Extraction Time (min)

518 754 749 530 506 550

1115 1102 1106 1098 1099 1130

100 100 100 100 200 100

30 30 30 30 15 30

RIGHT

AMS Lab Nos.

Oven Heat Time (hs)

BATCH

595

RADIOCARBON DATING BY STEP HEATING

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

Table I. Sample pretreatment and step heating procedures.

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

sites were selected for this research because their ages are well established and samples were readily available. A total of five sets of dates were run on bone and tooth enamel from the Dent site in Colorado. Stafford et al. (1991) produced a series of dates on various organic fractions of Dent mammoth bone, and believe its true age to be 10,810 ⫾ 40 B.P. This date is treated as the actual age of Dent bone in this study, but it should be noted that Taylor et al. (1996), opted for an age of 10,890 ⫾ 50 B.P. based on Stafford’s XAD hydrolysate fraction. One set of dates was run on mammoth tusk from the Murray Springs site in Arizona. Eight 14C dates on charcoal from the site, averaging 10,900 ⫾ 50 B.P. (Haynes, 1992), provide an actual age estimate for Murray Springs bone. Also, Clovis archaeological sites have been shown to date consistently within a narrow range of time from approximately 10,800 to 11,500 B.P. (Fiedel, 1999; Haynes, 1992; Taylor et al., 1996), providing further age control. Samples were subjected to various pretreatments to explore their effects on temperature-age spectra (Table I). Bone and tusk samples were cleaned of adhering sediment then coarsely ground and sieved to remove the fine fraction. Particle sizes used for dating ranged between 0.5 and 2 mm. A high speed rotary tool was used to clean enamel samples of adhering sediment and dentine. Enamel was pretreated in the form of flakes, generally ⬍ 5 mm in maximum dimension. Some samples were pretreated by heating in an electric furnace at 800⬚C. At this temperature, any organic matter is removed by reaction with O2 and secondary CaCO3 is broken down thermally, leaving the only source of carbon to be carbonate in the bone apatite. Other samples were subjected to wet preatments using combinations of acetic acid and NaOCl to remove contaminants. For CO2 extraction, approximately 1 g of pretreated sample was placed into a porcellanite boat within a quartz tube. From only 1 g of bone, it was possible to obtain at least seven dateable fractions of CO2 using the AMS technique. This contrasts to the 600 g used by Haas and Banewicz (1980) to recover four fractions dated by the conventional 14C method. The quartz tube with sample was placed onto a vacuum line for step heating and cryogenic separation of CO2. A tube furnace and thermocouple were used to control heating temperatures. Step heating procedures varied with respect to total temperature range, temperature intervals, and CO2 collection time (Table I). All dating was performed at the University of Arizona AMS 14C laboratory, and dates are reported as uncalibrated 14C ages with one sigma errors. RESULTS Dent Series 1 After mechanical cleaning, cortical bone from an unknown skeletal element was treated with 50% acetic acid for 24 h, and washed 5 times with distilled H2O. This was followed by several treatments of a 5:1 mixture 5% NaOCl and 50% acetic acid under weak vacuum for 48 h to remove residual organic matter. The addition of acetic acid results in the production of chlorine gas, a strong oxidizer, which speeds

596

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING Table II. Results of AMS 14C dating of Dent mammoth bone and tooth enamel by step heating. AMS Lab No.

Series

Material

Extraction Temp. (⬚C)

␦ 13C (PDB)

AA26641 AA26642 AA26643 AA26644 AA26645 AA26646 AA26647

Dent 1 Dent 1 Dent 1 Dent 1 Dent 1 Dent 1 Dent 1

Cortical bone Cortical bone Cortical bone Cortical bone Cortical bone Cortical bone Cortical bone

518 633 748 833 925 1018 1115

⫺14.9 ⫺11.0 ⫺6.4 ⫺4.5 ⫺7.9 ⫺8.6 ⫺9.4

8,700 9,770 9,750 9,850 10,200 10,230 10,080

⫾ 100 ⫾ 110 ⫾ 110 ⫾ 120 ⫾ 110 ⫾ 110 ⫾ 260

AA27984 AA27985 AA27986 AA27987

Dent 2 Dent 2 Dent 2 Dent 2

Enamel Enamel Enamel Enamel

754 882 994 1102

⫺8.0 ⫺3.9 ⫺4.2 ⫺4.7

3,365 8,305 8,385 8,525

⫾ 50 ⫾ 55 ⫾ 65 ⫾ 65

AA28261 AA28262 AA28263 AA28264

Dent 3 Dent 3 Dent 3 Dent 3

Enamel Enamel Enamel Enamel

749 908 1008 1106

⫺15.5 ⫺10.7 ⫺6.7 ⫺6.4

AA31318 AA31319 AA31320 AA31321 AA31322 AA31323 AA31324

Dent 4 Dent 4 Dent 4 Dent 4 Dent 4 Dent 4 Dent 4

Enamel Enamel Enamel Enamel Enamel Enamel Enamel

530 620 709 805 899 998 1098

⫺5.2 ⫺3.7 ⫺4.0 ⫺4.3 ⫺3.0 ⫺3.4 ⫺4.6

AA32634 AA32635 AA32636 AA32637

Dent 5 Dent 5 Dent 5 Dent 5

Cortical bone Cortical bone Cortical bone Cortical bone

506 700 910 1099

⫺10.5 ⫺4.5 ⫺4.3 ⫺8.4

Measured C Age ⫾ ␴

Expected C Age ⫾ ␴

% MC a

10,810 10,810 10,810 10,810 10,810 10,810 10,810

⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40

10.58% 4.87% 4.97% 4.47% 2.78% 2.64% 3.35%

10,810 10,810 10,810 10,810

⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40

53.74% 12.88% 12.41% 11.58%

Post-bomb 3,270 ⫾ 250 6,340 ⫾ 50 7,145 ⫾ 65

10,810 10,810 10,810 10,810

⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40

100% 54.76% 26.21% 20.35%

9,410 10,950 10,550 10,270 10,480 10,320 10,580

⫾ 150 ⫾ 230 ⫾ 110 ⫾ 110 ⫾ 100 ⫾ 100 ⫾ 110

10,810 10,810 10,810 10,810 10,810 10,810 10,810

⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40

6.70% 0% 1.16% 2.45% 1.48% 2.21% 1.02%

3,095 8,345 9,455 9,560

⫾ 65 ⫾ 65 ⫾ 80 ⫾ 75

10,810 10,810 10,810 10,810

⫾ 40 ⫾ 40 ⫾ 40 ⫾ 40

56.76% 12.65% 6.47% 5.92%

14

14

Percentage equivalent modern carbon contamination calculated as: % MC ⫽ (AMeas ⫺ AExp)/(13.56 ⫺ AExp), A ⫽ 13.56e ⫺ ␭t. a

the reaction. Informal experiments suggested that this mixture is more effective than pure NaOCl for bringing humic acids into solution. Finally, the sample was washed 5 times with distilled H2O and placed into an electric over at 95⬚C until dry. Step heating used seven 100⬚C temperature intervals ranging from approximately 400 to 1100⬚C. CO2 was collected for 30 min at each step. The dates produced ranged from 8700 ⫾ 100 to 10,230 ⫾ 110 B.P. (Table II; Figure 1), but none of the fractions were successful at replicating the true age of the bone. As expected, however, there is a clear trend of increasing age with greater temperature of extraction. The final fraction, collected at 1115⬚C, is slightly younger (10,080 ⫾ 260 B.P.) than the two previous samples, but its large standard deviation suggests that this is probably the result of measurement error. The general trend of increasing age with temperature does suggest contaminant carbonates are re-

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

597

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

Figure 1. Step heating radiocarbon dates on mammoth bone and tooth enamel from the Dent Clovis site, Colorado.

leased from the bone at lower temperatures, and with higher extraction temperature, the CO2 released is increasingly dominated by carbon indigenous to the bone. Dent Series 2 To eliminate the possibility of contamination by acetic acid or by exchange with dissolved carbonate in solution during pretreatment, the next sample was pretreated by oven heating. Also, mammoth tooth enamel was used instead of bone. For pretreatment, 1 g of clean tooth enamel was placed in an electric furnace at 800⬚C for 1 h. Step heating was performed as in Dent series 1, except only four temperature intervals of approximately 100⬚C were used ranging from 754 to 1102⬚C. Presumably, any carbon that could have been released at lower temperatures had been lost during pretreatment. The four dates produced ranged from 3365 ⫾ 50 to 8525 ⫾ 65 B.P. and increased with greater extraction temperature (Table II; Figure 1). Because enamel is expected to yield better results than bone, it was surprising that these dates were consistently younger than those produced in the previous series. A color change was noted in the enamel during step heating, which suggested that perhaps not all organic matter had been removed by pretreatment and that incomplete removal of organic matter in pretreatment could result in contamination of CO2 samples, even though there is no direct source of oxygen in the vacuum line for oxidation. Thus, the next experiment repeated the previous pretreatment, but extended the duration of oven heating to ensure complete removal of organic matter.

598

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING

Dent Series 3 After mechanical cleaning, 1 g of enamel was placed in an oven at 800⬚C for 12 h. Step heating again used four temperature intervals of approximately 100⬚C ranging from 749⬚C to 1106⬚C. The first CO2 fraction collected yielded a post-bomb age, and the remaining three increased in age with increased extraction temperature. Unfortunately, all yielded ages younger than comparable temperature fractions of the previous series (Table II; Figure 1). The post-bomb age and the general decrease in age seen in all temperature fractions suggested that pretreatment by oven heating was in fact causing exchange with atmospheric CO2, thus decreasing radiocarbon ages. Apparently, atmospheric exchange affected all temperature fractions suggesting that no fraction is totally immune to exchange, but that lower temperature fractions are more likely to exchange carbon than higher temperature fractions, at least under these conditions. A further complication introduced by atmospheric exchange of carbon by bone apatite at high temperatures is the possibility that the step heating process could result in mixing between low and high temperature fractions as the sample is exposed to CO2 released from itself during extraction. However, these effects should be minimized in the vacuum line by low pressures and short duration of exposure. Dent Series 4 Experience with series 2 and 3 strongly suggested that pretreatment by oven heating should be abandoned. For series 4, the pretreatment regime from series 1 was repeated with two modifications. First, tooth enamel was used instead of bone, and, second, all solution transfers were performed under a nitrogen atmosphere in a nitrogen tent to eliminate the possibility of atmospheric exchange. The sample only had contact with atmospheric CO2 during final oven drying at 95⬚C. The step heating procedure from series 1 was also employed. CO2 fractions were collected at seven 100⬚C temperature intervals from 530 to 1109⬚C. The results of series 4 were encouraging. Dates ranged from 9410 ⫾ 150 to 10,950 ⫾ 230 B.P. (Table II; Figure 1). The youngest date produced was at the lowest temperature interval (530⬚C). Surprisingly, the next fraction collected at 617⬚C was the oldest and fell well within the range of the actual age at 10,950 ⫾ 230 B.P. The remaining higher temperature fractions formed a plateau ranging between 10,270 ⫾ 110 and 10,580 ⫾ 110 B.P. At all temperature intervals, radiocarbon ages exceeded those of all previous series. Even though, one fraction did produce the expected age of the sample, it is possible this is due to measurement error. At 2 standard deviations, the 617⬚C sample overlaps with all higher temperature samples and given its position in the series at a relatively low extraction temperature, it is possible that the age of the sample, if measured more precisely, may be significantly younger. Dent Series 5 A final series of dates was run to address the possibility of contamination by acetic acid and of exchange during step heating. Ground and sieved cortical bone,

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

599

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

as in series 1, was pretreated with 5% NaOCl for 48 h under weak vacuum, washed 5 times with distilled H2O, and oven dried at 95⬚C. No acetic acid was used at any stage in pretreatment. It was therefore likely that secondary CaCO3 remained in the bone during step heating. However, any carbon it could contribute to CO2 samples should have been removed by 800⬚C (Haas and Banewicz, 1980:540). The step heating procedure was modified to minimize the time the sample was exposed to CO2 released at lower temperatures using only four 200⬚C temperature intervals from 506 to 1099⬚C and collecting CO2 was collected for only 15 min at each temperature interval. Series 5 showed intermediate results with dates ranging between 3095 ⫾ 65 and 9560 ⫾ 75 (Table II; Figure 1). The use of a strict NaOCl pretreatment with an abbreviated step heating regime resulted in dates significantly younger than those of series 1, which also used cortical bone. The discrepancy between series 1 and 5 is most likely explained by the use of acetic acid in pretreatment in the former. Acetic acid treatment is known to remove secondary CaCO3 from bone, but there is some evidence that it may also remove some carbonate from the apatite as well in that it is known to cause isotopic shifts in apatite stable carbon isotopes (Koch et al., 1997). This does not preclude the possibility that acetic acid pretreatment does contribute contamination to radiocarbon samples, but it suggests that it does more good than harm. The abbreviated step heating procedure used could contribute to the age discrepancy between series 1 and 5, but this seems very unlikely as Haas and Banewicz (1980:540) demonstrated that the rate of temperature increase has very little affect on the rate of sample weight loss when heating bone. Murray Springs Mammoth tusk from the Murray Springs site was selected for experimentation because previous work by Stafford et al. (1991) demonstrated that the organic material present in bone from this site is a very poor prospect for dating because its amino acid composition is largely noncollagenous. This provided an opportunity to test the Haas and Banewicz technique in a case where reliable dating of bone organics is not possible. Bone preservation from the Murray Springs site is in general very poor. In fact, the tusk that was used for dating was so poorly preserved that it crumbled out of its cast when excavators attempted to remove it from the site, and it currently exists as a few thousand tusk fragments. Proboscidean tusk is composed primarily of dentine which is intermediate to enamel and cortical bone in density and crystallinity (Fisher, 1986:347; Lyman, 1994:79), and therefore should be intermediate in its susceptibility to contamination. Pretreatment of the tusk sample was identical to that used above for Dent series 1. The tusk was coarsely ground and sieved to remove the fine fraction, followed by an acetic acid wash, and completed with a NaOCl/acetic acid wash. Step heating used seven 100⬚C temperature intervals ranging from 550 to 1130⬚C. CO2 was collected for 30 min at each interval. The dates produced ranged from 5040 ⫾ 55 to 10,815 ⫾ 55 B.P. (Table III; Figure

600

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING Table III. Results of AMS 14C dating of Murray Springs mammoth tusk dentine by step heating.a AMS Lab No.

Material

Extraction Temp. (⬚C)

␦ 13C (PDB)

AA33648 AA33649 AA33650 AA33651 AA33652 AA33653 AA33654

Tusk Tusk Tusk Tusk Tusk Tusk Tusk

550 642 747 811 898 1014 1130

⫺17.6 ⫺4.2 ⫺6.3 ⫺8.4 ⫺8.8 ⫺7.8 ⫺9.7

Measured C Age ⫾ ␴

14

5,040 10,625 10,815 10,655 10,620 10,710 10,605

⫾ 55 ⫾ 80 ⫾ 85 ⫾ 80 ⫾ 80 ⫾ 85 ⫾ 95

Expected C Age ⫾ ␴

% MC a

⫾ 50 ⫾ 50 ⫾ 50 ⫾ 50 ⫾ 50 ⫾ 50 ⫾ 50

37.24% 1.21% 0.37% 1.07% 1.23% 0.83% 1.30%

14

10,900 10,900 10,900 10,900 10,900 10,900 10,900

Percentage equivalent modern carbon contamination calculated as % MC ⫽ (AMeas ⫺ AExp)/(13.56 ⫺ AExp), A ⫽ 13.56e ⫺ ␭t. a

2). The temperature-age spectrum is distinctively step-shaped with a relatively young 14C age at 550⬚C (5040 ⫾ 55 B.P.), and dates plateauing after the first temperature interval. Dates between 642 and 1130⬚C vary slightly ranging from 10,605 ⫾ 95 to 10,815 ⫾ 55 B. P. and all overlap within 2 standard deviations. None of the average dates fall within the actual age range, estimated from an average of eight charcoal dates from the site, but the 747⬚C fraction overlaps it at 1 standard deviation. Also, two temperature fractions (747, 1014⬚C) fall within the of the charcoal

short standard

Figure 2. Step heating radiocarbon dates on mammoth tusk from Murray Springs, Arizona.

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

601

GEA(Wiley)

LEFT

BATCH

SUROVELL

total age range dates from the site, and the remainder overlap it within 2 standard deviations (Figure 6). The consistency of these dates with respect to extraction temperature and overlap with charcoal dates suggests that the true age of the bone may have been attained. If not, contamination is minimal ranging between 0.4% and 1.3% modern carbon content. These results demonstrate that reliable isotopic data can still be recovered from in bones in which the collagen has been seriously degraded and highlight the need for more experimentation with 14C dating of bone apatite carbonates. Summary and Analysis The choice of pretreatment methods is critical to the success of the step heating method as dating results varied widely depending upon the pretreatment used. The use of an acetic acid wash followed by a NaOCl/acetic acid wash gave the best results. Performing solution transfers under a nitrogen atmosphere may have improved dates as well, but more likely the success of Dent series 4 is due to the use of tooth enamel. The success of the dates from Murray Springs supports this contention as solution transfers were not performed under nitrogen. All of the series generally showed a trend of increasing age with increased extraction temperature. The lowest temperature fraction consistently produced the youngest dates and in three series (e.g., Dent series 2 and 4, Murray Springs) were significantly offset from the general trend of the other samples. This suggests that perhaps CO2 released at very low temperatures (⬍ 500⬚C) is derived in part from a highly exchangeable fraction of carbonate, perhaps that absorbed to crystal surfaces as postulated by Hedges et al. (1995:289). The general patterning seen in the Dent data suggests a system composed of a series of carbonate fractions that vary in their susceptibility to exchange, with no particular temperature fraction being totally immune. A simple model shows how temperature-age spectra may evolve with increased contamination (Figure 3). The model shows five temperature fractions, each being less susceptible to exchange than the previous lower temperature fraction. As a bone becomes increasingly contaminated, the slope of temperature-age spectra becomes increasingly steep. This occurs because low temperature fractions, being more susceptible to exchange, are contaminated at a greater rate than high temperature fractions. If such a model is accurate, it provides a convenient means of estimating the amount of contamination affecting a sample based on shape of the temperature-age spectrum produced. Spectra with a steep slope tend to produce 14C ages far from the actual age of the sample, while spectra with low slopes, tend to produce dates closer to the actual age of the sample. To test this model, slopes were determined for each series of dates using linear regression. The lowest temperature fractions were discarded as they tended to break the general trend of the remaining samples. Comparing slope to the equivalent percent “modern” carbon contamination in the best age produced from a given

602

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING

Figure 3. A hypothetical model of the evolution of temperature-age spectra with increasing contamination of the sample. Because each higher temperature fraction is less likely to undergo chemical exchange than the preceding fraction, the curves become steeper with greater contamination.

series demonstrates that spectra with steep slopes indeed tend to be characterized by greater contamination (Figure 4). Dent series 2 breaks from this trend, but this may be due to the abbreviated temperature range used for samples subjected to oven heating, because it has been subjected to atmospheric exchange. These data suggest that steepness of slope may be an effective means of independently evaluating the validity of radiocarbon ages, but further testing is needed. If, however, the relationship between slope and degree of contamination can be substantiated, it may eventually be possible to use this correlation to actually estimate the true age of a sample based on its response to increased contamination. Comparing the 14C dates produced in this study to those from previous studies using Dent mammoth bone shows the improvement gained in radiocarbon ages gained using the Haas and Banewicz technique (Figure 5). Dates from Dent series 4 showed tend to be slightly younger than the best dates obtained by Stafford et al. (1991) on various bone organic fractions. Although one date (625⬚C) does show significant overlap, this could be attributed to measurement error given its large standard deviation. Dating errors ranged between 0% and 2.5% modern carbon. There is also an improvement over Pigati’s (1996) dates on bulk enamel carbonates. For Murray Springs, step heating dates showed minor overlap with charcoal dates, but generally were younger than the estimated age of the site on the order of 0.4 – 1.3% modern contamination (Figure 6). Also, step heating dates consistently showed less contamination than bulk enamel carbonate dates (Pigati, 1996).

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

603

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

Figure 4. Slope versus percent modern carbon contamination for the Dent and Murray Springs series. Percent modern carbon contamination is based on the oldest date from each of the series: (ⵧ) pretreated by oven-heating; (●) not pretreated by oven heating.

CONCLUSIONS The step heating technique for 14C dating of bone apatite carbonate was able to produce the correct age on tooth enamel from the Dent site. Accurate ages may also have been produced for tusk dentine from Murray Springs. The slope of temperature-age spectra may provide a means of independently monitoring degree of contamination and validity of radiocarbon ages produced by step heating, but more work is needed to substantiate the relationship between the shape of step heating age spectra and level of sample contamination. The step heating method displays improvement over dates on bulk enamel carbonates, as errors were limited to ⬍ 2.5% modern carbon contamination for both sites. The success of the Haas and Banewicz technique is highly dependent on method of pretreatment. Dating results indicate that pretreatment with acetic acid followed by a 5:1 NaOCl/acetic acid mixture minimizes dating errors. It is possible that improved pretreatment could reduce errors further. Finally, the results of dating Murray Springs mammoth tusk indicate that useful radioisotopic data may be produced from bone in which the organic matter has degraded to the point of being undateable. These preliminary findings are encouraging and suggest that with further work, reliable radiocarbon dating of bone apatite may one day be a reality. On the other hand, until there is a reliable independent means of assessing the accuracy of radiocarbon dates on bone apatite, there is little reason to engage in dating of samples of unknown age.

604

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING

Figure 5. Comparison of step heating dates (Dent series 4) to bulk enamel carbonate (Pigati, 1996) and bone organic dates (Stafford et al., 1991) from Dent, Colorado in rank order: (●) bulk enamel carbonate; (䉫) step heating; (䉱) organics. Abbreviations: ENML, bulk enamel carbonates; ALA, alanine; ASP, aspartic acid; COLL, collagen; GEL, gelatin; GLU, glutamic acid; GLY, Glycine; HYP, hydroxyproline; THR, threonine; XAD-GEL, XAD-2-purified gelatin hydrolysate; XAD-HYD, XAD-2-purified collagen hydrolysate.

PROSPECTS FOR FURTHER RESEARCH Many avenues are available for improving upon the research presented in this article. Controlled experimentation with pretreatment and step heating regimes is badly needed. For example, the use of acetic acid seemed to improve 14C dates, and it is possible that the use of a stronger acid could improve dates further. Also, the possibility remains that oven drying at 95⬚C could have contributed some contamination to samples by atmospheric exchange. To alleviate this problem, samples could be dried by lyophilization or in a vacuum oven. The effects of varying step heating temperatures, increments, and extraction times remain unclear, something which future work could certainly address. Finally, further research could substantiate or refute the hypothesized relationship between the slope of temperatureage spectra and degree of sample contamination, and, if nothing else, further address the problem of independently assessing the accuracy of radiocarbon dates on bone apatite carbonate.

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

605

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL

Figure 6. Comparison of step heating dates to bulk enamel carbonate (Pigati, 1996) and charcoal dates (Haynes, 1992) from Murray Springs, Arizona in rank order (550⬚C step heating fraction omitted). (●) bulk enamel carbonate; (䉫) step heating; (— ) charcoal. Abbreviations: ENML, bulk enamel carbonates; CH, charcoal.

This research could not have been performed without the help of many people. C. Vance Haynes, Jr. was instrumental in providing samples and advice, but more importantly, he opened many doors that I could not have opened on my own. Many people at the University of Arizona AMS 14C facility were incredibly generous with their time and resources, in particular, Doug Donahue and Tim Jull. I am grateful to Christopher Eastoe, Austin Long, and Jay Quade for providing lab space, time, and supplies. Many other people have provided useful insights and advice, including Jeff Brantingham, Andrea Freeman, Herbert Haas, Vance Holliday, David Killick, Kris Kerry, Steve Kuhn, Mary Stiner, and Erv Taylor. Finally, thanks to Nicole Waguespack for making a valiant effort to translate my troubled writing into a form somewhat easier to read.

REFERENCES Berger, R. (1964). Radiocarbon dating of bone and shell from their organic components. Science, 144, 999– 1001. Fiedel, S.J. (1999). Older than we thought: Implications of corrected dates for Paleoindians. American Antiquity, 64, 95– 115. Fisher, D.C. (1986). Mastodont procurement by Paleoindians of the Great Lakes region: Hunting or scavenging. In M.H. Nitecki & D.V. Nitecki (Eds)., The evolution of human hunting (pp. 309– 421). New York: Plenum Press. Haas, H., & Banewicz, J. (1980). Radiocarbon dating of bone apatite using thermal release of CO2. Radiocarbon, 22, 537– 544. Hassan, A.A. (1976). Geochemical and mineralogical studies on bone material and their implications for radiocarbon dating, unpublished doctoral dissertation. Dallas: Southern Methodist University.

606

VOL 15, NO. 6

short standard

GEA(Wiley)

RIGHT

BATCH

RADIOCARBON DATING BY STEP HEATING Hassan, A.A., Termine, J.D., & Haynes, C.V., Jr. (1977). Mineralogical studies on bone apatite and their implications for radiocarbon dating. Radiocarbon 19, 364– 374. Haynes, C.V., Jr. (1967). Bone organic matter and radiocarbon dating. In Radiocarbon dating and methods of low level counting (pp. 163– 168). Vienna: IAEA. Haynes, C.V., Jr. (1968). Radiocarbon: Analysis of inorganic carbon on fossil bone and enamel. Science, 161, 687– 688. Haynes, C.V., Jr. (1992). Contributions of radiocarbon dating to the geochronology of the peopling of the New World. In R.E. Taylor, A. Long, & S. Kra (Eds)., Radiocarbon after four decades (pp. 355– 374). New York: Springer-Verlag. Hedges, R.E.M., Lee-Thorp, J.A., & Tuross, N.C. (1995). Is tooth enamel carbonate a suitable material for radiocarbon dating? Radiocarbon, 37, 285– 290. Hedges, R.E.M., & Millard, A.R. (1995). Bones and groundwater: Towards the modeling of diagenetic processes. Journal of Archaeological Science, 22, 155– 164. Holliday, V.T., Johnson, E., & Stafford, T.W., Jr. (1999). AMS radiocarbon dating of the type Plainview and Firstview (Paleoindian) assemblages: The agony and ecstasy. American Antiquity, 64, 444– 454. Irving, W.N., & Harington, C.R. (1973). Upper Pleistocene radiocarbon-dated artifacts from the Northern Yukon. Science, 179, 335– 340. Koch, P.L., Tuross, N.C., & Fogel, M.L. (1997). The effects of sample treatment and diagenesis on the isotopic integrity of carbonate in biogenic hydroxylapatite. Journal of Archaeological Science, 24, 417– 429. Lee-Thorpe, J.A., & van der Merwe, N.J. (1991). Aspects of the chemistry of modern and fossil biological apatites. Journal of Archaeological Science, 18, 343– 354. LeGeros, R.Z., Trautz, O.R., Klein, E., & LeGeros, J.P. (1969). Two types of carbonate substitution in the apatite structure. Experientia, 25, 5– 7. Libby, W.F. (1955). Radiocarbon dating. Chicago: The University of Chicago Press. Lyman, R.L. (1994). Vertebrate taphonomy. Cambridge: Cambridge University Press. McDougall, I., & Harrison, T.M. (1999). Geochronology and thermochronology by the 40Ar/39Ar method. New York: Oxford University Press. Nelson, D.E., Morlan, R.E., Vogel, J.S., Southon, J.R., & Harington, C.R. (1986). New radiocarbon dates on artifacts from the northern Yukon Territory: Holocene not upper Pleistocene in age. Science, 232, 749– 751. Person, A., Bocherens, H., Mariotti, A., & Renard, M. (1996). Diagenetic evolution and experimental heating of bone phosphate. Palaeogeography, Palaeoclimatology, Palaeoecology, 126, 135– 149. Pigati, J.S. (1996). Carbon isotopes in enamel bio-apatite: Applications to paleodietary reconstruction and geochronology, unpublished Master’s Thesis. Tucson: University of Arizona. Poyart, C.F., Fre´minet, A., & Bursaux, E. (1975). The exchange of bone CO2 in vivo. Respiration Physiology, 25, 101– 107. Quade, J., Cerling, T.E., Barry, J.C., Morgan, M.E., Pilbeam, D.R., Chivas, A.R., Lee-Thorp, J.A., & van der Merwe, N.J. (1992). A 16-Ma record of paleodiet using carbon and oxygen isotopes in fossil teeth from Pakistan. Chemical Geology(Isotope Geoscience Section), 94, 183– 192. Rey, C., Collins, B., Goehl, T., Dickson, R., & Glimscher, M.J. (1989). The carbonate environment in bone mineral: A resolution-enhanced Fourier transform infrared spectroscopy study. Calcified Tissue International, 45, 157– 164. Saliege, J.F., Person, A., & Paris, F. (1995). Preservation of 13C/12C original ratio and 14C dating of the mineral fraction of human bones From Saharan tombs. Journal of Archaeological Science, 22, 301– 312. Shipman, P., Foster, G., & Schoeninger, M. (1984). Burnt bones and teeth: An experimental study of color, morphology, crystal structure and shrinkage. Journal of Archaeological Science, 11, 307– 325. Sillen, A. (1989). Diagenesis of the inorganic phase of cortical bone. In T.D. Price (Ed.), The chemistry of prehistoric human bone (pp. 221– 229). Cambridge: Cambridge University Press. Stafford, T.W., Jr., Duhamel, R.C., Haynes, C.V., Jr., & Brendel, K. (1982). Isolation of proline and hydroxyproline from fossil bone. Life Sciences, 31, 931-938.

GEOARCHAEOLOGY: AN INTERNATIONAL JOURNAL

607

short standard

GEA(Wiley)

LEFT

BATCH

SUROVELL Stafford, T.W., Jr., Jull, A.J.T., Brendel, K., Duhamel, R.C., & Donahue, D. (1987). Study of bone radiocarbon dating accuracy at the University of Arizona NSF Accelerator Facility for Radioisotope Analysis. Radiocarbon, 29, 24– 44. Stafford, T.W., Jr., Brendel, K., & Duhamel, R.C. (1988). Radiocarbon, 13C and 15N analysis of fossil bone: Removal of humates with XAD-2 resin. Geochimica et Cosmochimica Acta, 52, 2257– 2267. Stafford, T.W., Jr., Hare, P.E., Currie, L., Jull, A.J.T., & Donahue, D.J. (1991). Accelerator radiocarbon dating at the molecular level. Journal of Archaeological Science, 18, 35– 72. Stiner, M.C., Kuhn, S.L., Weiner, S.L., & Bar-Yosef, O. (1995). Differential burning, recrystallization, and fragmentation of archaeological bone. Journal of Archaeological Science, 22, 223– 237. Tamers, M.A., & Pearson, F.J. (1965). Validity of radiocarbon dates on bone. Nature, 208, 1053– 1055. Taylor, R.E. (1987). Radiocarbon dating: An archaeological perspective. Orlando: Academic Press. Taylor, R.E. (1994). Radiocarbon dating of bone using accelerator mass spectrometry: Current discussions and future directions. In R. Bonnichsen & D.G. Steele (Eds.), Method and theory for investigating the peopling of the Americas (pp. 27– 44). Corvallis, Oregon: Center for the Study of the First Americans. Taylor, R.E., Haynes, C.V., Jr., & Stuiver, M. (1996). Clovis and Folsom age estimates: Stratigraphic context and radiocarbon calibration. Antiquity, 70, 515– 525. Weiner, S., & Bar-Yosef, O. (1990). States of preservation of bones from prehistoric sites in the Near East: A survey. Journal of Archaeological Science, 17, 187– 196.

Received February 1, 2000 Accepted for publication March 1, 2000

short standard

608

VOL 15, NO. 6