A re-analysis of electron spin resonance dating results associated with the Petralona hominid

Rainer Grün Quaternary Dating Research Centre, ANH, RSPAS, Australian National University, Canberra ACT 0200, Australia Received 24 February 1995 Revision received 3 November 1995 and accepted 22 November 1995 Keywords: ESR dating, Petralona, chronology.

A re-analysis of electron spin resonance dating results associated with the Petralona hominid Two electron spin resonance (ESR) dating studies of the late 1970s and early 1980s on speleothems associated with the Petralona hominid cranium yielded age estimates of 350–700 ka and 200 ka, respectively. These dating results are re-assessed in view of more than a decade of progress in ESR dating. The re-assessed values are in reasonable agreement with some independent U-series results on the same material, suggesting an age of 150–250 ka for the speleothems bracketing the cranium. ? 1996 Academic Press Limited

Journal of Human Evolution (1996) 30, 227–241

Introduction The age of the Petralona hominid cranium is of great importance for the understanding of human evolution. The circumstances of the discovery of the Petralona skull and the subsequent radiometric dating illustrates the range of problems that are associated with the establishment of reliable chronologies. A skull was found in 1960 in a cave near Petralona, Northern Greece, in a chamber that is referred to as the Mausoleum. The floor of this chamber is covered with a thick flowstone layer. It is not clear whether the skull was found on the flowstone, or whether it was lying on the underlying sediment and the flowstone layer was precipitated afterwards (Cook et al., 1982). It was claimed (e.g., Poulianos, 1980) that the rest of the complete skeleton was found beneath the flowstone but was subsequently lost. This claim has been disputed by others (e.g., Xirotiris et al., 1982). Hennig et al. (1981) established, using neutron activation analysis of trace elements, that a brown calcite crust partly covering the skull, correlated to the uppermost layer of the thick, otherwise white flowstone and that the underlying white flowstone layers showed significantly different elemental compositions. Along with a complete lack of sediment adhering to the skull, this evidence makes it most probable that the specimen lay on the thick white flowstone rather than embedded in sediments found below this layer and that both the white flowstone and the skull were subsequently encrusted with a thin brown calcite layer (Cook et al., 1982). The anthropological classification of the skull went through several stages. It was first claimed that the skull was a variant of the Neanderthal type (e.g., Kokkoros & Kanellis, 1960; Brose & Wolpoff, 1971), whereas others saw a resemblance to Rhodesian Man (e.g., Howells, 1967) or modern Man (Poulianos, 1972). A multivariate study by Stringer (1974) concluded that the skull itself shows anthropological features closely related to the specimen from Broken Hill (Zimbabwe) or Djebel Irhoud, with some Homo erectus features. Stringer (1974), therefore, claimed that the specimen must be older than Upper Pleistocene. Murril (1981) was not quite sure whether the skull should be classified as H. erectus petralonensis/rhodesiensis or Homo sapiens petralonensis/rhodesiensis, pending the (unknown) age of the specimen. Further studies came to the conclusion that the specimen represents the most primitive grade of the species H. sapiens (Stringer, 1980, 1981; Wolpoff, 1980). The newest classification seems to group the Petralona specimen together with other European and African fossils, such as Arago and Broken Hill, into a distinct Eurafrican Middle Pleistocene archaic H. sapiens species, Homo heidelbergensis (Hublin, 1985; Stringer 1985, 1988, 1990). 0047–2484/96/030227+15 $18.00/0

? 1996 Academic Press Limited

228

. ¨ 

The assessment of the associated faunal remains is also ambiguous. Sickenberg (1964) first described the fauna associated with the skull as Eemian, but revised this later to Biharian [=Middle Pleistocene: Sickenberg (1971)]. Kurtén & Poulianos (1977) classified the fauna associated with the hominid remains as Cromerian with an age of around 700,000 years. However, in a review of European hominid sites, Cook et al. (1982) (1) disputed that any of the faunal remains found in the Petralona Cave could be clearly associated with the hominid specimen and (2) pointed out that the faunal chronology, which led to the old age assessment, was fraught with serious contradictions. Numerous dating studies have been carried out on the site including U-series analyses of stalagmitic material underlying the skull (but supposedly covering the skeleton) and on the calcite crust of the skull itself. Additionally, many dates were produced on samples with dubious relationships to the hominid cranium from other sites in the cave. These results are not further discussed in this paper. There are three groups of samples that are supposedly directly related to the cranium (see also Table 1): (1) the red-brown calcite encrustation of the skull; (2) a thin laminated reddish-brown layer on top of the flowstone covering the floor of the Mausoleum; and (3) thick white flowstone below this reddish layer. There is speculation that the skull is encrusted with two distinct calcite coatings: an upper reddish-brown layer and a lower almost colourless, pale brownish-grey one (Liritzis, 1982). However, Hennig et al. (1981) observed that the reddish-brown layer was directly deposited onto the cranium. Shen & Yokoyama (1984) described two samples directly above and below the presumed emplacement of the cranium. Both samples showed a reddish non-laminated top layer followed by a well laminated brownish-white calcite. Between these layers, smaller pieces of fossil bones were found. If this observation can be related in any way to the encrustation of the cranium it would seem that the reddish-brown layer was indeed directly deposited onto the skull. The dating results carried out on the three groups of samples are summarized in Table 1. Some of the sample descriptions are not detailed enough to allow a precise correlation of the samples of the various studies. Therefore, the correlation of the samples to the ones of Hennig et al. (1981) have to be regarded as tentative. U-series analyses of the encrustation seemed difficult because of ‘‘bone contamination’’ (sample P-6 of Liritzis, 1982) or a very low 230Th/232U ratio of about 1 (sample CC of Shen + 46 , which he & Yokoyama, 1984). Liritzis reported a non-corrected U-series result of 84"32 corrected into an age of about 130–150 ka. Shen and Yokoyama (1984) did not carry out any age calculation on their sample, however, the un-corrected isotopic mean values would result in an age of 82 ka. The thin laminated reddish-brown layer on top of the flowstone has been correlated with the skull encrustation by neutron activation analysis (Hennig et al., 1981). Cook et al. (1982) report an unpublished U-series result by Latham and Schwarcz of 166&25 ka. Later, Latham & Schwarcz (1992) gave a preferred U-series age range of 160–200 ka for this sample (75GR4/1). Liritzis (1982) reported corrected U-series age estimates from the same layer in the range of 138–175 ka (sample P-12) whereas Shen & Yokoyama (1984) gave an age estimate of + 211 371"74 ka for this layer (PL3) and >350 ka for an equivalent sample on the cave wall just above the skull (PL2). The contradictory U-series results may, on the one hand, cast some doubts on the correlations of the samples, but, on the other hand, this may also be a sign of open system behaviour of uranium. The upper part of the white flowstone under the cranium, which gives the older age bracket, + 122 (GR4 top: has been dated to 137–165 ka (sample P-13 of Liritzis, 1980) and 273"68

 -   Table 1

229

Age assessments of calcite samples directly related to the Petralona cranium: U-series analyses and ESR analyses

U-series analysis Sample No.

Age (ka)

Comments and tentative correlation to samples of Hennig et al. (1981)

Liritzis (1980, 1982) Schwarcz et al. (1980) (all U-series results corrected for detrital Th) +19 P-12 154"17 Reddish layer on top of floor flowstone, corresponding to (b) +15 P-13 150"13 White sample just below P-12, corresponding to (d) +46 P-6 84"32 From burial site of skull, corrected to 130 and 150 ka due to bone contamination, corresponding to (a) +122 GR-4 top 273"68 Flowstone underlying cranium, corresponding to (b) or (d) +£ GR-4-2 lower 451"130 Flowstone underlying cranium, corresponding to (e) +84 GR-4-3 lower 283"57 Flowstone underlying cranium HU-38 >450 Flowstone underlying cranium Shen & Yokoyama (1984) +33 PL1 209"24 Stalactite from roof, no relationship to cranium PL2 >350 Salagmitic veil on wall just above cranium, mixture to two layers corresponding to (b) and (d) +211 PL3 371" Reddish layer below cranium, corresponding to (b) 74 PL4 >350 Floor of Mausoleum, corresponding to (d); calcite on cranium, no calculation CC carried out because of clay contamination. Average U-isotopic measurements would result in about 82 ka, corresponding to (a) Latham & Schwarcz (1992) + 23 75GR4/1 top 231"19 Flowstone underlying cranium, corresponding to (b) + 26 75GR4/1 1·3 178"23 Flowstone underlying cranium, corresponding to (d) + 27 75GR4/1 1·5 166"24 Flowstone underlying cranium, corresponding to (d) + 29 75GR4/1 1·7 153"25 Flowstone underlying cranium, corresponding to (d) 75GR4 4/2 middle >350 Flowstone underlying cranium, corresponding to (e) 75GR4 4/2 bottom >350 Flowstone underlying cranium, corresponding to (e) ESR Analyses Sample number

Age (ka)

Dose (Gy)

Dose rate (µGy/a)

Comments

#4

330

660

2000

#5

340

670

2000

Carbonate above skull, may correspond to (b) or (d) Above skeleton, may correspond to (d) or (e)

(a) (b) (c) (d) (e)

198&40 195&40 127&35 198&50 650&280

417&42 419&42 271&27 386&39 817&123

2110&210 2114&210 2130&360 1950&300 1260&350

Ikeya (1980)

Hennig et al. (1981) Calcite encrusting skull Top reddish calcite on floor Bone fragments of skull White calcite 3–4 mm below (b) Calcite 30–40 mm below (b)

Schwarcz et al., 1980). Other U-series results on deeper samples of the white flowstone are all infinite (Schwarcz et al., 1980; Hennig, 1979; Hennig et al., 1981; Shen & Yokoyama, 1984; Latham & Schwarcz, 1992). ESR dating was first carried out by Ikeya (1980). Two samples relate to the cranium: #4 is described as carbonate above the skull, which may either relate to the encrustation or to a sample from the cave wall (similar to sample PL2 of Shen & Yokoyama, 1984), and #5 originated from ‘‘above skeleton’’ (i.e. the underlying flowstone). Because of uncertainties in the assessment of the external dose rate, Ikeya (1980) favoured first an age of 350 ka for both samples whereas his co-worker (Poulianos, 1980, 1981) pronounced an age of 700 ka, in line

. ¨ 

230

with the faunal chronological assessment. The interpretation of the ESR results was criticized by Wintle & Jacobs (1982) because of the large uncertainties in the dose-rate estimation. Hennig et al. (1981), also using ESR, obtained an age estimate of 200&40 ka on the calcite encrustation, the reddish-brown top layer of the flowstone, as well as the white flowstone immediately under the reddish layer. A sample from the lower section of the white flowstone gave an ESR age estimate of 650&280 ka. These age estimates were later disputed and defended in an exchange of letters in the journal Nature (Poulianos et al., 1982). In order to clarify the re-assessment of the ESR analysis, a short introduction into the basics of ESR dating is required. ESR dating of speleothems The dating of speleothems was the first application of ESR dating (Ikeya, 1975). It was thought for some time that this application was particularly promising because relatively small samples could be dated very rapidly and this was an advantage over radiocarbon and U-series dating. Moreover, a dating range of a few thousand to about a million years seemed possible. However, few systematic ESR dating studies have been carried out on speleothems, and it has not been possible to prove the reliability of ESR age estimates beyond the U-series limit of about 350,000 years. Mass spectrometric U-series dating now allows high-precision dating on very small samples to about 400,000 years (see Edwards et al., 1987a,b), although such analyses have not yet been carried out on Petralona samples. ESR is a dating method that is based on the measurement of trapped charges that accumulate in minerals over time (for recent reviews see Grün, 1989a,b; Ikeya, 1993). Figure 1 shows the basic principle: radioactive rays eject negatively charged electrons from atoms in the ground state (valence band). The electrons are transferred to a higher energy state, the so-called conduction band, leaving positively charged holes near the valence band. After a short time of diffusion, most electrons recombine with holes and the mineral returns to its original state. However, all natural minerals contain imperfections, such as lattice defects, interstitial atoms etc, which can trap electrons when they fall back from the conduction band. These trapped electrons can be measured by ESR spectroscopy, giving rise to a characteristic ESR lines (Figure 2). The intensity of the ESR line is proportional to the number of trapped electrons and the number of trapped electrons in turn results from three parameters: (1) the strength of the radioactivity (dose rate), (2) the number of traps (sensitivity) and (3) the duration of radiation exposure (age). The simple scheme of Figure 1 also shows the two major limitations of ESR dating: (1) saturation: when all traps are filled, any additional radiation cannot result in a further increase in the number of trapped electrons. (2) thermal stability: electrons have only a limited probability of staying in the traps. After a certain period of time (the so-called thermal mean life, ô), 63% of an original population of trapped electrons will have left the traps and will have recombined with holes. The effect of recombination or fading can only be neglected if the mean life is at least ten-times larger than the age of the sample. Because the mean life is dependent on the nature of the paramagnetic centre and the ambient temperature, there is no general upper dating limit for ESR dating and each type of material has to be evaluated individually. ESR age estimates are derived from the following formula: Age=

Accumulated dose (DE) Dose rate (D ~)

 -  

231

Figure 1. Trapping of electrons: the basis for ESR dating. An insulating mineral has two energy levels which electrons may occupy. The lower energy level (valence band) is separated from the higher energy level (conduction band) by a so-called forbidden zone. When a mineral is formed or reset, all electrons are in the ground state. Ionising radiation ejects negatively charged electrons (@) from atoms. The electrons are transferred to the conduction band and positively charged holes (%) are left behind near the valence band. After a short time of diffusion most of the electrons recombine with the holes. Some electrons can be trapped by impurities (electron traps) in the crystal lattice. These electrons can be directly measured by electron spin resonance spectroscopy (see Figure 2). Ea is the activation energy or trap depth, which controls the thermal stability of the trapped electrons.

As can be seen in the left part of Figure 3, natural radiation enhances the ESR line. The intensity value of the natural sample corresponds to a dose value, DE, that the sample has received since it was formed. The measurement of DE is the actual ESR part of the dating procedure. This value is determined by the additive dose method (right part of Figure 3, see below). The dose rate, D ~ , is derived from the chemical analysis of the radioactive elements (U, Th, and K; other radioactive elements are usually negligible) in the sample and its surroundings. The isotopic concentrations are converted into dose rates using published tables (e.g. Nambi & Aitken, 1986). The determination of the radioactivity that influences the sample is rather complex and has to be carefully evaluated. As can be seen from the equation above, two parameters have to be determined in order to ~. estimate an ESR age: the accumulated dose, DE, and the dose rate, D

. ¨ 

ESR-intensity [a.u.]

232

2.005

2.0

1.995 g-value

Figure 2. ESR signals of a young speleothem sample (top) and an old one (below). The upper spectrum is amplified relative to the lower spectrum by a factor of ten. The ESR signal intensity is a measure of the number of trapped electrons. The different ESR signals result from separate, characteristic paramagnetic centres. It can be immediately seen that the intensity of the ESR lines is a measure of the age.

Determination of the accumulated dose, DE For the determination of DE, the ESR response to radioactive irradiation must be known and mathematically expressed. For this purpose, aliquots of the sample are irradiated with a gamma source (e.g., 60Co). Many hospitals and chemistry departments have such gamma sources for a variety of applications (e.g. sterilization or cancer treatment). During radiation an amount of energy is transferred to the sample and the unit for this energy dose is Gray (Gy). [The old unit which is still often used is rad (100 rad=1 Gy)] In the laboratory, a piece of speleothem (between 2 and 5 g) is separated and thin surface layers of about 50–100 µm are removed (to eliminate the volume irradiated by alpha rays). The sample is ground and sieved for homogenization and about ten equal quantities (aliquots) are weighed out (about 150 mg) and irradiated by a calibrated gamma source. About 2 weeks after the irradiation the sample is measured with an ESR spectrometer. An ESR signal used for dating should have the following properties: (a) the ESR signal is zero when the sample is formed; (b) the signal intensity grows with added dose; (c) the lifetime of the signals must be at least one order of magnitude higher than the age of the sample;

 -  

233

450 400

Natural radiation

ESR intensity (a.u.)

350

Artificial irradiation

300 250 200 150 100

DE

50 0 –100

Natural ESR intensity –50

0

50

100

150

200

250

300

Radiation dose (Gy) Figure 3. Determination of the accumulated dose, DE. Natural radiation generated the measured natural ESR intensity in the sample. Aliquots of the sample are successively irradiated with increasing gamma doses and the associated ESR intensities are plotted vs. the laboratory doses. The DE value results from the extrapolation to zero ESR intensity. The fitting is normally carried out using a single saturation function, with weighting inversely proportional to the square of the measured ESR intensity and using an error assessment as suggested by Brumby (1992).

(d) the number of traps is constant or changes in a predictable manner; recrystallization, crystal growth or phase transitions must not have occurred; (e) the signals should not show any fading which is not not related to the thermal stability; (f) the ESR signal should not be influenced by sample preparation (grinding, exposure to laboratory light). The signal intensity is plotted vs. the calibrated laboratory gamma doses (additive dose method). The extrapolation to zero ESR intensity yields the DE value on the intersection with the X-axis (Figure 3). At the time of formation, the sample is not supposed to contain any trapped electrons and, hence, no ESR intensity (this assumption may not be correct for some materials such as calcretes or spring deposited travertines: see e.g., Grün et al., 1988; Radtke et al., 1988). Although this extrapolation seems rather straightforward, there have recently been intense discussions about the correct procedures for DE determination and the assessment of the errors involved. Detailed simulations of dose–response curves (Grün & Brumby, 1994) have demonstrated that most ESR data sets are best fitted by a single saturating exponential function, weighting that is inversely proportional to the square of the measured ESR intensity and an error calculation that has been suggested by Brumby (1992). The signal at g=2·0005 (see Figure 2 and Figure 4: upper spectrum) is most commonly used for dating speleothems. The signal also occurs in other secondary carbonates such as aragonitic mollusc shells and corals. The signal is isotropic and Debuyst et al. (1990, 1991) proposed that this signal is associated with a rotating CO " 2 radical. The actual thermal stability (in nature) of this paramagnetic centre is probably lower than suggested by Hennig & Grün (1983) who found a value of ô in the range of 107 years at 10)C. Rossi & Poupeau (1989) found that the thermal stability may vary from sample to sample. It was suggested that the signal may

234

. ¨ 

be composed of a stable and an unstable component (Mudelsee, 1990; Mudelsee et al., 1992; Barabas et al., 1992a). For a comprehensive review of the signal at g=2·0005, the reader is referred to Barabas et al. (1992a,b). The broad line shown in Figure 4 (lower spectrum) has also been used for dating. The use of this ESR signal may lead to severe age over-estimations (Grün, 1985, 1989a) especially when dating spring deposited travertines (Grün et al., 1988). In some cases, however, reliable age estimates of speleothems can be obtained using this broad line (Wieser et al., 1985). Hennig and Grün (1983) showed in their Figure 20 that the DE estimation (using the signal at g=2·0005: Hennig, pers. comm.) was critically dependent on the microwave power used in the measurement. Values of 24–49 Gy had been obtained for microwave powers in the range of 1·5–45 mW. The interpretation of Hennig & Grün (1983) was, that saturation effects were responsible for this trend and that the results obtained from the low microwave power should be the most reliable. Unfortunately, this hypothesis is not correct. The signal at g=2·0005 sometimes has a satellite, a thermally unstable peak at g=2·0018. When using low microwave powers and large modulation amplitudes (e.g., 2 Gpp) as routinely used by Hennig in the early eighties, both signals occur as a single line and the resulting DE may grossly underestimate the correct dose value (Grün & DeCanniere, 1984). The satellite peak, however, is suppressed by higher microwave powers (e.g., Molodkov, 1988; Barabas et al., 1992a,b). Therefore, DE measurements at high microwave powers are more likely to be correct when using large modulation amplitudes. Determination of the Dose Rate, D ~ The strength of the radioactive field which irradiates the sample is determined by the concentration of radioactive elements in the sample and its surroundings plus a component from cosmic rays. In ESR studies, only the U- and Th-decay chains and the 40K-decay are of relevance (as only a minor contribution comes from 87Rb in the sediment). There are three ionizing rays that are emitted from the radioactive elements (the ranges are given for a material with a density of about 2·5 g/cm3): (1) alpha particles (He nuclei) have only a very short range of about 20 µm because of their large size. They are not as efficient in producing ESR intensity as beta and gamma rays, therefore an alpha efficiency, k value (which is usually in the range of 0·05–0·3), has to be determined (see Aitken, 1985, 1990). (2) Beta particles (electrons) have a range of about 2 mm. (3) Gamma rays have a range of about 30 cm. Because the concentrations of radioactive elements in the sample and its surroundings are usually very different, it is necessary to determine the internal dose rate separately from the external dose rate. Additionally, the effect of cosmic rays has to be considered. The cosmic dose rate is dependent on the geographic latitude, the altitude and the thickness of the covering sediments. The cosmic dose rate is about 300 µGy/a at sea level and decreases with depth below ground (Prescott & Stephan, 1982; Prescott & Hutton, 1988). For speleothems the following dose rate parameters have to be determined: U-concentration, 234U/238U ratio, alpha efficiency, gamma dose rate, and thickness for gamma attenuation (if sample is bigger than 3–5 cm). The internal dose rate of speleothems is nearly exclusively controlled by the uranium content. The uranium concentrations are usually low and the contribution from the internal dose rate is very small. Lyons and Brennan (1991) determined a value of 0·052&0·026 for the alpha efficiency of speleothem calcite.

 -  

235

100 000 (a) 50 000 g = 2.0005 0

ESR intensity (a.u.)

–50 000 Natural sample Irradiated sample –100 000 3442 10 000 (b)

3447

3452

3457

3462

3467

3472

3477

3482

3472

3477

3482

5000 g = 2.0040 0

–5000 Natural sample Irradiated sample –10 000 3442

3447

3452

3457

3462

3467

Magnetic field (G) Figure 4. ESR spectra of speleothems. (a) shows the ESR line around g=2·0005 which has been successfully used for dating. The broad line around g=2·0040, shown in (b) may lead to ESR age over-estimations.

In a cave, the external dose rate is generated by a multitude of sources such as the cave walls, speleothems, clay layers on the floor, etc. It is, therefore, necessary to determine the external gamma dose rate in situ with a thermoluminescence dosimeter or a portable gamma spectrometer. Any assumptions about the magnitude of the gamma dose rate may lead to erroneous results. The cosmic dose rate is usually negligible in deeper caves. Re-analysis of ESR age estimates from Petralona The data available from the early ESR dating studies are here re-analysed to assess the reliability of these results. As discussed above, it is very important to select an ESR signal that is likely to give reliable results. Ikeya shows only one spectrum of a sample (which is not in any relationship to the cranium) from Petralona (Figure 1 in Ikeya, 1977), a single line that is attributed to CO33" . Unfortunately, this statement cannot be used to identify the nature of this specific line. The width of the line seems to correspond to the broad line in Figure 4(b). The use of this line may lead to age over-estimations. Ikeya did not give the microwave power or the modulation amplitude used in the 1980 study but reports 4 Gpp for the 1977 results. Should the measured ESR signal be composed of the g=2·0005 and the g=2·0018 signals, any DE values would present underestimations.

. ¨ 

236

600

(a)

800

DE = 41.7 ± 9.5 IMAX = 1941 ± 4017 D0 = 1258 ± 2870

500 400

DE = 368.9 ± 152.4 IMAX = 20700 ± 692000 D0 = 41787 ± 1419420

700 600 500 400

300

300

200

200 100

100

600

(b)

500 400

800

DE = 368.8 ± 124.5 IMAX = 3746 ± 20517 D0 = 5082 ± 30366

700 600

0 10 0 20 0 30 0 40 0 50 0 60 0 70 0 80 0

–5 0 –4 0 0 –3 0 0 –2 0 0 –1 0 00

30 0

20 0

10 0

0

00 –1

00 –2

–3

00

0

00

0 –4

ESR intensity (a.u.)

(c)

(d) DE = 349.9 ± 34.6 IMAX = 2417 ± 1168 D0 = 3026 ± 1803

500 400

300

300

200

200 100

100 0 –4 0 0 –3 0 0 –2 0 0 –1 0 00 0 10 0 20 0 30 0 40 0 50 0 60 0 70 0 80 0

–5

0 30

0 20

0 10

0

00 –1

00 –2

00

0 –3

–4

00

0

Dose (Gy) Figure 5. Dose response data from Ikeya (1980): diagrams a and b and Hennig et al. (1981): diagrams c and d. Exponential fitting (solid line) results in somewhat smaller DE values than linear fitting (dotted line) which was applied in the original studies. The curve parameters and errors refer to the results of the exponential fitting.

In line with all other ESR studies at that time, only a few aliquots of each sample were used for the estimation of the DE value, and linear fitting was applied to the data points without any error calculation. Newer views of data processing include exponential fitting, weighting of ESR intensity inversely proportional to the squared ESR intensity, and analytical models for error calculation (Grün & Brumby, 1994). In order to estimate the possible random and systematic errors, the two data sets of Figure 1 in Ikeya (1980) were analysed according to Grün and Brumby (1994) [see Figure 5(a) and 5(b)]. Instead of a dose value of 44 Gy given for the lower-data set, the re-analysis results in 41·7&9·5 Gy. The upper-data set leads to 368·8&124·5 Gy (there is no dose value cited for these data; linear fitting results in 389 Gy). The results indicate that the cited DE values may be associated with systematic errors in the region of 10%, and random errors of about 30%. The assessment of the reliability of the dose rate values is more difficult. Ikeya (1980) refers to Ge(Li) data presented in Ikeya (1977) where a stalagmite from Petralona (not related to the cranium at all) was analysed by gamma-ray spectroscopy. The elemental results of 0·572 ppm U and 0·145 ppm Th along with an alpha efficiency of 0·1 are used to postulate a dose rate of 1990&170 µGy/a (3040 µGy/a were calculated for an alpha-efficiency of 0·2). Ikeya (1980) mentions an alpha-efficiency of 0·4 which would, according to these calculations, result in a dose rate of more than 5000 µGy/a. However, Ikeya’s dose rate calculations are not reproducible. When using the conversion factors of Nambi and Aitken (1986), the elemental analysis would lead to a total dose rate of 258 µGy/a for k=0·1 and 767 µGy/a for k=0·4

 -  

237

(excluding the gamma component). One measurement of the environmental gamma dose rate in Petralona cave with a CaSO4 dosimeter yielded a value of 870&200 µGy/a (Ikeya, 1978). Despite this confusion, the dose rates of 1000 and 2000 µGy/a are regarded as tentative (Ikeya, 1980). If we use a total dose rate of 1500&500 µGy/a (which seems reasonable considering the reconstructed internal dose rates and measured external dose rate) and taking into account possible systematic and random errors of the DE estimation, the ESR age estimates of samples 4 and 5 would both yield age estimates of about 400&180 ka. The true error is probably larger due to possible additional systematic errors in the selection of measurement conditions and nature of the ESR signal used for DE estimation. If a total dose rate of 2200&400 µGy/a was used, as reconstructed from Hennig et al. (1981), see below, the age estimate would be 273&96 ka. It seems reasonable to conclude that the published data of Ikeya (1980) do not allow a more precise age estimate for the two samples that relate to the cranium than the range of about 175–580 ka. Hennig et al. (1981) did not state the modulation amplitude or the g-value of the signal used for DE estimation. The microwave power of 2 mW is not sufficient to suppress the unstable signal at g=2·0018. If the spectra were measured with a large modulation amplitude (e.g., 2 Gpp) and the signal for dating was around g=2·0005, the DE values may present underestimations. If, on the other hand, the signal at g=2·0040 was used (the high alpha-efficiency of 0·3 may give some indication for this), the DE-values may have been overestimated. Hennig et al. (1981) used linear extrapolation for DE estimations. The error of 10% assigned to all DE values resulted from two repeated DE determinations of samples (a) and (b). On re-analysis of their Figure 2 (under the same conditions as for Ikeya’s data sets) one obtains DE values for sample (a) of 369&152 Gy and 349&35 Gy instead of 396&40 and 438&44 Gy, respectively [see Figure 5(c) and 5(d)]. A systematic underestimation in the 10–20% range and random errors in the 10–30% range may also be assumed for samples (b)–(e). The dose rate calculations of Hennig et al. (1981) are straightforward. All samples were analysed for U, Th, and K, the alpha efficiency was determined (0·30&0·06) and the external dose rate was measured in situ (1900&200 µGy/a). There is a minor mistake in the dose rate calculations in so far as the internal dose rates for Th and K also contain the gamma component (this is already part of the measured gamma dose rate). If the gamma component is taken out of the total dose rate, the value of 2004&210 µGy/a is well within the range of 2110&210 µGy/a given by Hennig et al. (1981). It may also be argued that the alpha dose rate from 232Th is not negligible (Hennig argued that the Th was bound in agglomerated clay clusters). Considering the full alpha dose rate from 232Th, the total dose rate would result in 2367&210 µGy/a. Therefore, an averaged dose rate of 2200&400 µGy/a would cover any uncertainty. Hennig et al. (1981) also analysed a sample of the cranium itself [sample (c)] reporting an age of 127&35 ka. This age assessment has to be regarded as a minimum age estimate, because re-crystallization processes in bones lead to systematic ESR age underestimations (Grün & Schwarcz, 1987). Hennig et al. (1981) also suspected that recrystallization was responsible for the young age estimate of sample (c). Discussion Sample (c) of Hennig et al. (1981) gives a minimum age for the cranium of 126&35 ka. The crust on the cranium [sample (a)] gives revised age estimates of 165&74 and 155&32 ka

238

. ¨ 

instead of 198&40 ka. The alpha-spectroscopic U-series results of the crust were problematic because of contaminations (Liritzis, 1982, Shen & Yokoyama, 1984). A corrected U-series age estimate was in the range of 130–150 ka (Liritzis, 1982). The re-evaluated ESR results are also in remarkably good agreement with the U-series results of the reddish crust at the top of the floor flowstone of 138–175 ka (Liritzis, 1982) and in reasonable agreement with the value of + 23 ka (Latham & Schwarcz, 1992). It may be assumed that sample (d) (198&50 ka after 231"19 Hennig et al., 1981), which is supposed to give the older age bracket, will give similar results to sample (a), again in astonishingly good agreement with sample P-13 (137–165 ka) of Liritzis (1982) and the analyses of Latham & Schwarcz (1992), which are all in the range of about 150–180 ka. However, the U-series results of Shen & Yokoyama (1984) on other reddish samples suggest a much older age for supposedly equivalent samples. One may conclude that the differences between the re-analysed ESR age estimations of Ikeya (1980) (about 175–580 ka) and Hennig et al. (1981) (about 90–240 ka) are actually not significant. On the other hand, the discrepancies between the ESR as well as the U-series laboratories may point to the problem that the samples do not correlate directly. It may well be that there are several reddish-brownish layers with distinct ages. The published isotopic data may give some indication for this suspicion: all analyses of the cranium crust show very high Th concentrations [Th(a)=1·84 ppm; Th(P-6)=1·70 ppm and Th(CC)~1·5 ppm]; the uranium seems very variable (in the range of 0·066–8 ppm). The Th concentrations of the upper reddish layer of the travertine floor also show relatively high values [Th(b)=1·39 ppm, Th(P-12)=0·91 ppm, Th(P-13)=0·73 ppm]. However, the Th concentrations of the reddish-wall samples are very low [Th(PL2) and Th(PL3) <0·2 ppm]. If sample #5 of Ikeya (1980) actually relates to sample (d) of Hennig et al. (1981), the age estimates are basically the same. The same applies to sample #4: if it originates from the wall above the skull, it gives very comparable results to PL2 and PL3 of Shen and Yokoyama (1984). Considering all dating results, it seems clear that: (1) the cranium has a minimum ESR age of 126&35 ka, based on the analysis of the bone; (2) the crust covering the skull is well within the range of U-series dating, but it may be difficult to determine precise age assessments due to contamination problems; (3) all U-series studies on the white layer immediately under the reddish top layer of the floor travertine give finite U-series age assessments; (4) all U-series results on the deeper white travertine are infinite (i.e., >300 ka). The results imply that the age of the Petralona hominid is well within the range of U-series dating, i.e., it is younger than 350 ka. If it is accepted that the age of the hominid is bracketed by the crust on the cranium (giving the lowest possible age) and the thin white layer immediately under the reddish-brown top layer of the floor travertine (giving the upper age), the most likely age of the Petralona hominid is in the range of 150–250 ka. Although it may now be possible to obtain more precise ESR age estimations (by using more aliquots and a larger dosage range for DE estimation and the measurement of all parameters), one may doubt whether a higher accuracy will be obtained. The final age assessment of the Petralona hominid specimen may depend on high precision mass spectrometric U-series dating of the crust as suggested by Schwarcz (1992). However, the alpha-spectroscopic U-series measurements imply that precise age assessments may be difficult due to contamination problems. Samples of the crust could be traced through contacting N. J. Xirotiris, who removed calcite from the skull (Xirotiris, pers. comm. in 1982) although Schwarcz (pers. comm. in 1995) has been told that the samples were thrown away. An alternative lies in direct gamma spectrometric dating of the hominid specimen (see e.g., Yokoyama et al., 1988).

 -  

239

One may argue that the morphology of the Petralona cranium suggests a much older age than 150–250 ka because there are similarities with Arago (dated to 400–700 ka, see DeLumley & Labeyrie, 1981) and Broken Hill (hitherto undated). However, in this context the present author would like to point out that morphological features do not necessarily represent chronology. For example, modern humans have lived at about 100–120 ka in South Africa and the Levant (see Grün & Stringer, 1991 and references therein), but in Europe they occurred at about 40 ka, which is a factor of three, later. Neanderthals, on the other hand, showing more archaic features, survived for a long time after the occurrence of the first modern humans, even in the same geographical area. A similar case can be made for the survival of H. erectus in east Asia. This is to say that the evolution of morphological features is not a straight chronological line, but may vary to a great extent in different macro, and probably micro, geographical areas.

Conclusions The re-assessment of published ESR data show that the errors associated with the age estimates are considerably larger than those cited in the original papers. Additionally, the procedures used in the original studies may have lead to age over-estimations. The re-evaluated ESR age estimates are in good agreement with those U-series dating results that can be directly correlated to the samples. All dating studies taken together suggest an age of about 150–250 ka for the Petralona hominid. Acknowledgements I wish to thank C. B. Stringer, Natural History Museum, London, H. P. Schwarcz, McMaster University, and two unknown referees for constructive comments, J. Papps and F. M. Grün, Canberra, for corrections on the manuscript. References Aitken, M. J. (1985). Thermoluminescence Dating. London, New York: Academic Press. Aitken, M. J. (1990). Science-based Dating in Archaeology. London: Longman. Barabas, M., Mudelsee, M., Bach, A., Walther, R. & Mangini, A. (1992a). General properties of the paramagnetic centre at g=2·0006 in carbonates. Quat. Sci. Rev. 11, 165–171. Barabas, M., Mudelsee, M., Walther, R. & Mangini, A. (1992b). Dose response and thermal behaviour of the ESR signal at g=2·0006 in carbonates. Quat. Sci. Rev. 11, 173–179. Brose, D. S. & Wolpoff, M. H. (1971). Early Upper Paleolithic man and late Middle Paleolithic tools. Am. Anthropol. 73, 1156–1194. Brumby, S. (1992). Regression analysis of ESR/TL dose–response data. Nuclear Tracks 20, 595–599. Cook, J., Stringer, C. B., Currant, A. P., Schwarcz, H. P. & Wintle, A. G. (1982). A review of the chronology of the European Middle Pleistocene hominid record. Yearb. of Phys. Anthropol. 25, 19–65. Debuyst, R., Bidiamambu, M. & Dejehet, F. (1990). Diverse CO " 2 radicals in ã- and á-irradiated synthetic calcite. Bull. Soc. Chim. Belg. 99, 535–541. Debuyst, R., Bidiamambu, M. & Dejehet, F. (1991). An EPR study of ã- and á-irradiated synthetic powdered calcite labelled with carbon 13. Nuclear Tracks 18, 193–201. DeLumley, H. & Labeyrie, J. (eds) (1981). Absolute Dating and Isotope Analysis in Prehistory—Methods and Limits, Paris: Proceedings, Pretirage. Edwards, R. L., Chen, J. H., Ku, T. L. & Wasserburg, G. J. (1987a). Precise timing of the last interglacial period from mass spectrometric determination of thorium-230 in corals. Science 236, 1547–1553. Edwards, R. L., Chen, J. H. & Wasserburg, G. J. (1987b). 238U-234U-230Th-232Th systematics and the precise measurement of time over the past 500,000 years. Earth Planetary Sci. Lett. 81, 175–192. Grün, R. (1985). Beiträge zur ESR-Datierung. Sonderveröffentlichungen Geol. Inst. Univ. Köln 59, 1–157.

240

. ¨ 

Grün, R. (1989a). Electron spin resonance (ESR) dating. Quat. Int. 1, 65–109. Grün, R. (1989b). Die ESR-Altersbestimmungsmethode. Berlin, Heidelberg: Springer. Grün, R. & Brumby, S. (1994). The assessment of errors in the past radiation doses extrapolated from ESR/TL dose response data. Radiat. Measure. 23, 307–315. Grün, R. & DeCanniere, P. (1984). ESR dating: problems encountered in the evaluation of the naturally accumulated dose (AD) of secondary carbonates. J. Radioanal. Nuclear Chem., Lett. 85, 213–226. Grün, R. & Schwarcz, H. P. (1987). Some remarks on ‘‘ESR dating of bones’’. Ancient TL 5, 1–9. Grün, R., Schwarcz, H. P., Ford, D. C. & Hentzsch, B. (1988). ESR dating of spring deposited travertines. Quat. Sci. Rev. 7, 429–432. Grün, R. & Stringer, C. B. (1991). ESR dating and the evolution of modern humans. Archaeometry 33, 153–199. Hennig, G. J. (1979). Beiträge zur Th-230/U-234-Altersbestimmung von Höhlensintern sowie ein Vergleich der erzielten Ergebnisse mit denen anderer Absolutdatierungsmethoden. PhD Dissertation, Köln, Universität zu Köln, 173+XVII pp. Hennig, G. J. & Grün, R. (1983). ESR dating in Quaternary geology. Quat. Sci. Rev. 2, 157–238. Hennig, G. J., Herr, W., Weber, E. & Xirotiris, N. I. (1981). ESR-dating of the fossil hominid cranium from Petralona Cave, Greece. Nature 292, 533–536. Howells, W. W. (1967). Mankind in the Making. London: Penguin. Hublin, J. J. (1985). Human fossils from the North African middle Pleistocene and the origin of Homo sapiens. In (E. Delson, Ed.) Ancestors: The hard Evidence, pp. 283–288. New York: Alan R. Liss. Ikeya, M. (1975). Dating a stalactite by electron paramagnetic resonance. Nature 255, 48–50. Ikeya, M. (1977). Electron spin resonance dating and fission track detection of Petralona stalagmite. Anthropos (Athens) 4, 152–166. Ikeya, M. (1978). Natural radiation at Petralona cave measured with thermoluminescence dosimeter of CaSO4(Tm). Anthropos (Athens) 5, 54–58. Ikeya, M. (1980). ESR dating of carbonates at Petralona Cave. Anthropos (Athens) 7, 143–150. Ikeya, M. (1993). New Applications of Electron Spin Resonance—Dating, Dosimetry and Microscopy. Singapore, New Jersey, London, Hong Kong: World Scientific. Kokkoros, P. & Kanellis, A. (1960). Découverte d’une crâne d’homme paléolithique dans la péninsule Chalcidique. L’Anthropologie 64, 438–446. Kurtén, B. & Poulianos, A. N. (1977). New stratigraphic and faunal material form Petralona cave, with special reference to the Carnivora. Anthropos (Athens) 4, 47–130. Latham, A. G. & Schwarcz (1992). The Petralona hominid site: uranium-series re-analysis of ‘‘Layer 10’’ calcite and associated palaeomagnetic analyses. Archaeometry 34, 135–140. Liritzis, Y. (1980). 230Th/234U dating of spelaeothems in Petralona. Anthropos (Athens) 7, 215–241. Liritzis. Y. (1982). Petralona Cave dating controversy. Nature 299, 280–181. Lyons, R. G. & Brennan, B. J. (1991). Alpha/gamma effectiveness ratios of calcite speleothem. Nuclear Tracks 18, 223–227. Molodkov, A. (1988). ESR dating of Quaternary shells: recent advances. Quat. Sci. Rev. 7, 477–484. Mudelsee, M. (1990). ESR-Datierung karbonathaltiger, quartärer Tiefseesedimente. Unpublished Diplomarbeit, Heidelberg, Universität Heidelberg. Mudelsee, M., Barabas, M. & Mangini, A. (1992). ESR dating of the Quaternary deep-sea sediment core RC17-177. Quat. Sci. Rev. 11, 181–189. Murril, R. I. (1981). Petralona Man. Springfield, Illinois: Thomas. Nambi, K. S. V. & Aitken, M. J. (1986). Annual dose conversion factors for TL and ESR dating. Archaeometry 28, 202–205. Poulianos, A. N. (1972). Some sapiens features of the Petralona skull. In (F. Bordes, Ed.) The origin of Homo sapiens. Paris: UNESCO. Poulianos, A. N. (1980). The postcranial skeleton of the Archanthropus europaeus petraloniensis. Anthropos (Athens) 7, 13–29. Poulianos, A. N. (1981). Pre-sapiens man in Greece. Curr. Anthropol. 22, 287–288. Poulianos, A. N., Liritzis, Y., Ikeya, M., Hennig, G. J., Herr, W., Weber, E. & Xirotiris, N. I. (1982). Petralona cave dating controversy. Nature 299, 280–282. Prescott, J. R. & Hutton, J. T. (1988). Cosmic ray and gamma ray dosimetry for TL and ESR. Nuclear Tracks Rad. Measure. 14, 223–227. Prescott, J. R. & Stephan, L. G. (1982). The contribution of cosmic radiation to the environmental dose for thermoluminescence dating—latitude, altitude and depth dependencies. PACT 6, 17–25. Radtke, U., Brückner, H., Mangini, A. & Hausmann, R. (1988). Problems encountered with absolute dating (U-series, ESR) of Spanish calcretes. Quat. Sci. Rev. 7, 439–445. Rossi, A. M. & Poupeau, G. (1989). Radiation-induced paramagnetic species in natural calcite speleothems. Appl. Rad. Isotopes 40, 1133–1137. Schwarcz, H. P. (1992). Uranium series dating in paleoanthropology. Evol. Anthropol. 1, 56–62. Schwarcz, H. P., Liritzis, Y. & Dixon, A. (1980). Absolute dating of travertines from Petralona Cave, KhalkidikiGreece. Anthropos (Athens) 7, 152–167.

 -  

241

Shen, G. & Yokoyama, Y. (1984). Th-230/U-234 dating of Petralona spelaeothems. Anthropos 11, 23–32. Sickenberg, O. (1964). Die Säugetierfauna der Höhle Petralona bei Thessaloniki. Geol. Geophys. Res. (Institute for Geology and Subsurface Research, Athens) 9, 1–16. Sickenberg, O. (1971). Revision der Wirbeltierfauna der Höhle Petralona (Griechisch Mazedonien). An. Géol. Pays Helléniques 23, 230–264. Stringer, C. B. (1974). A multivariate study of the Petralona skull. J. human Evol. 3, 397–404. Stringer, C. B. (1980). The phylogenetic position of the Petralona cranium. Anthropos (Athens) 7, 81–95. Stringer, C. B. (1981). The dating of European Middle Pleistocene hominids and the existence of Homo erectus in Europe. Anthropologie (Brno) 19, 3–14. Stringer, C. B. (1985). Middle Pleistocene hominid variability and the origin of Late Pleistocene humans. In (E. Delson, Ed.) Ancestors: The hard Evidence, pp. 289–295. New York: Alan R. Liss. Stringer, C. B. (1988). Archaic Homo sapiens/Petralona. In (I. Tattershall, E. Delson & J. Van Couvering, Eds) Encyclopedia of Human Evolution and Prehistory, pp. 49–54 and 447. New York, London: Garland. Stringer, C. B. (1990). In (G. P. Rightmire, Ed.) The Evolution of Homo erectus. Cambridge: Cambridge University Press. Wieser, A., Göksu, H. Y. & Regulla, D. F. (1985). Characteristics of gamma-induced ESR spectra in various calcites. Nuclear Tracks 10, 831–836. Wintle, A. G. & Jacobs, J. A. (1982). A critical review of the dating evidence for Petralona cave. J. Archaeol. Sci. 9, 39–47. Wolpoff, M. H. (1980). Cranial remains of the Middle Pleistocene European hominids. J. human Evol. 9, 357–364. Yokoyama, J., Falgueres, C. & Bibron, R. (1988). Direct dating of Neanderthalian remains and animal bones by the non-destructive gamma-ray spectrometry: comparison with other methods. L’Homme Néandertal 1, 135–141. Xirotiris, N. J., Henke, W. & Hennig, G. J. (1982). Die phylogenetische Stellung des Petralona Schädels auf Grund computertomographischer Analysen und der absoluten Datierung mit der ESR-Methode. Humanbiol. Budapestinensis 9, 89–94.