The first inter-laboratory ESR comparison project phase II: Evaluation of equivalent doses (ED) of calcites

NuN. Tracks, Vol. 10. Nos 4-6. pp. 945-952, 1985

0191-278X/8553.00+0.00 Pergamon Press Ltd.

Printed in Great Britain

THE FIRST I N T E R - L A B O R A T O R Y ESR C O M P A R I S O N PROJECT PHASE II: E V A L U A T I O N OF E Q U I V A L E N T DOSES (ED) OF CALCITES G. J. HENNIG, M. A. GEYH Nieders/ichsisches Landesamt fiir Bodenforschung Alfred-Bentz-Haus, Stilleweg 2, D-3000 Hannover 51, F.R.G. and R. GRON* Geologisches Institut der Universit/it zu K61n Lehrstuhl fiir Eiszeitenforschung, Ziilpicher Str. 49, F.R.G. (Received 14 November 1984; in revised form 15 March 1985)

Abstract--In 1982, four homogenized calcite samples of differing ages and origin and one fossil bone sample were investigated by 16 laboratories engaged in ESR dating. The goal of phase II of this project was to determine the precision of ED evaluation that can be obtained with this method and to find sources of error (Hennig et aL, 1983). The data scatter widely which is partly due to improper recording and evaluation of the spectra. Apart from preliminary recommendations for a more reliable determination of equivalent doses, systematic studies to improve ED evaluation by ESR is highly recommended.

INTRODUCTION ELECTRON spin resonance (ESR) spectra of natural calcites generally exhibit two or more resonance lines (ESR signals) around the g value for the spin of the free electron: g = 2.0023. Additional ESR signals appear in the spectra of irradiated samples, hence the choice of a suitable ESR signal for the determination of equivalent doses (ED) is often problematic. ESR signals that must not be used for this purpose are (a) those occurring in modern samples, (b) those with a mean life of the electrons in the corresponding trap which is not lower than the sample age, (c) those with interference from other signals, and (d) those showing light sensitivity or abnormal fading effects. ESR signals have been frequently observed in natural calcites at eight different g values, which are given in Table 1 (Hennig and Grfin. 1983). Of these, only

those at g values of 2.0035 and 2.0007 seem to be suitable for ED evaluation. The signals at g values > 2.0040 generally seem to be due to organic radicals and are often detected in samples of recent material. The signal at g = 2.0022 is thermally unstable, the one at g = 2.0002 is due to surface defects introduced by milling or grinding, and the one at g = 1.9996 is generally interfered with by Mn 2+, which is frequently present in the calcite lattice.

RESULTS OF THE C O M P A R I S O N P R O J E C T (PHASE II) Four calcite samples (TROM, PRI-8, HEGY, and PARG) and one bone sample were sent to 25 laboratories known to be engaged in ESR dating in 1982, 16 of which have submitted results mostly without details of measuring techniques and data evaluation. Additionally, three TL laboratories determined ED values by thermoluminescence. The results are summarized in Table 2 and shown in Fig. 1.

*Present address: Department of Geology, McMaster University, 1280 Main Street West. Hamilton, Ontario, Canada L8S 4M 1. 945

946

G. J. H E N N I G et al. Table 1. ESR signals frequently present in calcites Mean g-factor

•Linewidth at 9 GHz

Mean life at 1 0 ° C

2.0058 2.0042 2.0035 2.0028 2.0022 2.0007 2.0002 1.9996

1-2 4-6 ca. 0.6 ca. 3 ca. 1 ca. l ca. 1 1-2

106 106 0.02-4 × 106

Comments properties organic radicals? organic radicals light sensitive? only upon heating unstable suited for dating caused by grinding Mn -'+ interference

2-3 0.3-7 × 106

Table 2. ED results (Gy) of the inter-laboratory comparison project Lab. Code (coded) 4 5 6 7 9 10 12 13 14 16 21 22 24 25 26 27 TL-2 TL-7 TL-21

Sample TROM -420+70 55 + 10 -< 10 12 + 2 318 + 32 20 +'8 30 + 15 --30 22 7.7 + 12•1 29 12 4- 4 -->_ 300(?) (360°C)

Sample PRI-8 0 + 2.5 16__+ 14 8.04 + 0.09 < 10 25+--5 3.3 4-2 75 4- 8 45 + 62 -5+ 3 2 _ 0.4 <2 5 27.3 4- 8.5 1 1.6 + 0.3 ---

Sample HEGY 500 541 +95 49 +- 24 220 4- 30 515+_60 225 + 45 483 ~ 48 494 + 85 650 + 50 608 + 45 470 150 300 1030 +_.39 31 300 + 50 190-250 200 + 18

--

STATISTICAL EVALUATION Any g e o c h r o n o l o g y c o m p a r i s o n project must include a rigorous statistical t r e a t m e n t o f the data, but because everybody involved pays special attention to calibration and p r o p e r analysis performance, such tests c a n n o t simulate the everyday routine o f dating a sample o f u n k n o w n age. The 58 s u b m i t t e d E D values A~ (Table 2) contain four that represent detection limits and twelve were submitted w i t h o u t s t a n d a r d deviation o A , The other values were s u b m i t t e d with relative s t a n d a r d deviations which cover an unusually wide range o f 3 - 1 6 0 % , i n d e p e n d e n t o f the c o r r e s p o n d i n g absolute value o f ED. T w o explanations are possible: (1) A p r o c e d u r e for d e t e r m i n i n g actual s t a n d a r d

Sample PARG 25•7 _ 5 46+6 62 _ 2 36 + 4 25+ 10 18 + 5 64 + 6 49 + 50 -25 +- 27 < 10 20 32 144 _ 10 17 22 +_.2 -50 4- 3 20

155

deviations has not yet been developed for E D evaluation or is not generally known. W h e n this is the case, the data may be treated as if the precisions o f the different values are similar and then G a u s s i a n statistics may be used• (2) The s t a n d a r d deviations are correct a n d must be taken into account in any statistical evaluation• F o r the first case, the arithmetic m e a n M and the m e a n s t a n d a r d deviation o f a single m e a s u r e m e n t a M were calculated as follows: n

g

+-aM

= I

t/

+--I,,,1 n ~

~

'

(1)

For the second case, the weighted mean W and the mean s t a n d a r d deviation ~ W o f a single m e a s u r e m e n t

T H E F I R S T I N T E R - L A B O R A T O R Y ESR C O M P A R I S O N P R O J E C T

ICP-ESR: Equivotenf Doses

frequency distribution, ehi-square values (Z 2) were calculated using the following equation:

z 2 = V ' (,4,.-,.t

0

1

2"

3

t,

5

6

30

2b

o

60

8o.

"R0 *

TL -

(3)

"

The m a x i m u m Z 2 values for a test with a probability of 1% are as follows: 2 . Zmax' n u m b e r of dates:

i j HoY i I do

)2

oA-

~0 kr~

'R"J l

[!.ll

947

16o

kraal

I

I

dafa

FIo. 1. ICP-ESR: Equivalent Doses (krad; l krad = l0 Gy) submitted for the 4 calcite samples.

7 9 2 3

I1 13 17 19 20 22 23 4 5 7 8 9 10 11

The results of this statistical evaluation are compiled in Table 3. To check whether the wide scatter of the data was due to bias shifts, two normalized frequency distributions were calculated for the dates of each laboratory. In the first case, the ratios AJw and the standard deviations obtained from equation (1) were calculated. This procedure has the advantage that standard deviations need not be known. Assuming a precision of + 5% (Hennig and Griin, 1983), the result is 1 + 0.05. In the second case, we calculated oA values using equation (4). Standard deviations and X2 values were then calculated using equations (1) and (3), respectively.

with the weight 1 were calculated in the following manner:

A i -- w

6A~= ~

(4)

Wi

n

~i Ai/aA

w +_.ow=

1/oA~

±/El" -"

i 1/aA~

"

(2)

1

The calculation was repeated after deletion of subjectively selected outliers to obtain w _ aw. To check whether the dates for each sample belong to the same

A theoretical value of 0 + 1 is expected. The results of this bias test are compiled in Table 4. The relative standard deviations of the arithmetic means (equation (1)) are + 60%, independent of the absolute ED values (Table 3). The chi values show that the data for each sample do not belong to same population, indicating that various systematic errors

Table 3. Results of the statistical evaluation of the ED data (Gy) Topics Number of results representing detection limits without standard deviations Range of aA/A Samples

TROM

PRI8.

12 1 2 10-160

15 2 2 10-140

HEGY

PARG

16 15 0 I 5 3 4--50 3-100 HEGY 410 + 254

Total 58 4 12

TROM 87 __+142

P R 18 15 + 22

Number of outliers

15 __ 5 + 30 3

2.3 +0.8 +- 35 1

364 4-42 +- 12 2

41 +-4 + 10 3

W 4- O'W

13 4- 4

2.2 4- 0.7

392 + 48

26 + 4

+- #w/w (%) Chi square (number of data) Chi square without outliers

+ 30 143 (8) 4.2 (5)

± 30 162 (11) 79 (10)

+- !2 579 (11) 111 (9)

M±oM W+_-~W +aw/w(%)

PARG 42 + 33

+- 15 356 (10) 24 (7)

948

G.J.

H E N N I G et al.

Table 4. Results of a bias test for each laboratory: ~rA and 7~-~were calculated with the equations 4 and 3, respectively, n is the number of used data Laboratory code

A/w

4 5 6 7 9 10 12 13 14 16 21 22 24 25 26 27

0.75 + 1.14 (+ 1 5 2 % ) 10.7+ 18.4(+171°/o) 2.6+ 1,8(+69%) 2.2+2.5(+113%) 3.6+5.9(+162%) 0.92+0.42(+46%) 15.5+23.4(+151%) 6.3+ 11.2(+178%) 2.0+ 1.5(+75%) 1.5+_0.9(+60%) 0.83 +_0.22(+26%) 1.1 +0.9(+-83%) 1.6+- 1.0(+-63%) 5.3+_7.1(+_134%) 0.85+0.96(+_113%) 0.82+0.23(+_28%)

(some of which are discussed below) are a major part of the total error ( a M ) given in Table 3. The relative standard deviations from the weighted means (calculated without extreme outliers) range from 12 to 30%. Apart from the results of sample T R O M (without outliers), the corresponding dates do not belong to single frequency distributions. Hence, none of the calculated mean values A, IV, and w is representative of the actual ED values. The evaluation of the A~/w values for each laboratory (Table 4) does not yield evidence for bias shift since 12 of 16 laboratories show standard deviations for the normalized A / w that are larger than + 50%. Only two (Nos. 21 and 27) yielded + 25-30%. The mean aA values deviate considerably from 0 + 1 with the exception of the results from laboratories Nos 4, 10, 13, and 27. Taking into account the standard deviations, this ~rA test (equation 4) yielded a larger number of laboratories with apparently good work procedures compared to the A / w test. However, owing to the higher sensitivity of this test, bias shifts become obvious (Table 4, second column). To sum up, the best inter-laboratory reproducibility of ED determinations is +_ 25%, in most cases (14 of 16) even worse than + 50%. DISCUSSION Sample B O N E This sample was prepared from a large femur bone of a Pleistocene elephant. It was excavated from the

aA

,/.z

-0.9+0.9 0.8 +2.9+4.0 48.3 + 18.7+4.0 32.5 -3.2+6.3 39.l +6.5+3.5 25.0 -5.3+l.7 16.9 +26.9__+8.5 217.4 + 3 .2 + 1.0 2.9 +6.2+5.2 27.9 . . . . ----+5.7+3.5 23.9 +30.6+11.7 134.6 ---6.1 +2.0 11.6

n 2 4 4 2 3 4 4 4 2 --3 4 -4

Krefelder Terrasse, Kempen, FRG. All but three of the laboratories classified this sample as unsuited for deriving an ED. Consequently, the results of this sample are not discussed. Sample T R O M This calcite sample was collected at Mechernich-Breitenbenden, FRG, from a travertine deposit in a tunnel-type, brick aqueduct connecting springs in the Eifel area and the city of Cologne, The aqueduct was in use between AD 100 to AD 500. The internal dose rate was estimated t,o be 0.5 mGy yr -L on the basis of U, Th, and K contents of 0.39, 0.45 and 650 ppm respectively (determined by gamma-spectrometry), and an assumed alpha efficiency of k = 0.2. Applying a typical gamma dose rate of ca. 1.4 mGy yr- ~ for pottery and soil (Aitken, 1974), the total dose rate may be on the order of 1 . 9 m G y y r - L Consequently, the equivalent dose should be _< 4 Gy, even with a k value of 0.5. Nearly all of the laboratories obtained ED values greater than 4 Gy, for which several explanations may be given. At least one of the two ESR laboratories with ED values > 300 Gy used the presumed organic radical signal at g = 2.0050 (Fig. 2). Indeed, an apparent ED of about Gy is obtained when using this ESR signal. One laboratory reported an ED value of 300 Gy from the 360 C TL peak because there was no TL emission in the usual range of 230-270 C. This finding indicates that spurious TL from calcites at 360 C may be

T H E F I R S T I N T E R - L A B O R A T O R Y ESR C O M P A R I S O N PROJECT

Sample TROM 971.5MHz 2roW 0 5G~

/

nat.

~l

Mn2*~

i g= 2 0022

"2:°2 3650

31.70

949

Sample HEG Y This is a speleothem sample from the cave "Grosse Wiliheimshrhle" at Heggen-Finnentrop, F.R.G. It has a U-series age greater than the dating limit, i.e. > 300 kyr, and is of high purity. The ESR spectrum of the natural sample (Figs 4 and 5, top) shows a large signal at g = 2.0006, which seems to be suitable for an ED evaluation (Table 1). Of the four other lines (at g = 2 . 0 1 1 5 , 2.0023, 2.0002, and 1.9975), the first one has not yet been detected in other samples. The "grinding" signal at g = 2.0002 and the unstable signal at g = 2.0023 can be properly separated from the "dating" signal at g = 2.006 only when the ESR spectrum is recorded with a small modulation of, for example, 0.5 Gpp (Fig. 4, below and Fig. 5, center). If a modulation of, for example, 2 Gpp is used, the unstable and the stable ESR signals, are superposed (Fig. 5, below). In the latter case, the slope of the growth curve will be too steep and the resulting ED will be apparently too small. The diagrams of at least three laboratories show that instable peaks or the "grinding" peak were used for the ED

mognetic held[G] Sample PRI-8 9 71.2MHz 2mW

FIG. 2. ESR spectra of the natural and irradiated sample TROM.

o.5 G~

caused by organic radicals similar to those causing the ESR signal at g ~ 2.0050. Other laboratories may have used the instable ESR signal at g = 2.0023 and/or the surface defect signal (g = 2.0002), as indicated in one of the submitted spectra. Last, but not least, the high Mn 2+ concentration has certainly interfered with the relatively small signal at g = 2.0007.

Sample PRI-8 PRI-8 is a homogenized sample from a large post-glacial stalagmite in the cave "Prinzenhrhle" at Hemer-Sundwig, F.R.G., with a very low Mn 2+ content (Fig. 3). The U-series age of this stalagmite is 9 + 1 yr near the center and 3 _.+ 1 k yr near the surface. The internal dose rate is due to only 0.1 ppm U, and the external dose rate measured in the same cave system may be assumed to not exceed 0.5 mGy y r - ~. Using the maximum age of 9 k yr, the expected ED is less than 5 Gy. Indeed, 11 of the 15 laboratories reported values in this range, which is an encouraging result.

nat. {x /~)

•"

•

31.'60

~

j g=2.0001

31.80

magnetic field/G]

FIG. 3. ESR spectra of the natural and irradiated sample PRI-8.

950

G . J . H E N N I G et al. Sample HEGY 97{,1. MHz 2row

O.S G~

•gx20023 g=20f~5

*40krad

r=199?5 g=2.000~

31.60

ESR spectrum of the natural, non-irradiated sample (Fig. 7) has only one signal (g = 2.0038)., located between two of the forbidden transition lines from Mn -'+. The stability of this signal seems to differ greatly between samples of different origin and a light sensitivity has been observed at this trap level for most samples (Smith et aft., 1985; Hennig and Griin, 1983). The latter interference may be a reason for the scatter of the data (Fig. 1). The two TL results for this sample area also in disagreement (by a factor of 2.5). RECOMMENDATIONS The following recommendations for ESR dating work are derived from the investigation reported here, as well as from our own experience and the ESR literature, keeping well in mind that further improvements in the method are necessary. The g values mentioned have an uncertainty of about 0.0002. 1. For the dating of calcites, a suitable, stable ESR signal generally appears at g = 2.007. Mean lives of 0.3 x 106-7 x 106 have been determined for the trap

31.80 magnetic field[G~

FIG. 4. ESR spectra of the natural and irradiated sample HEGY.

evaluation. This explains the apparently low ED values of about 200 Gy, Four laboratories used the properly discriminated signal at g = 2.0006, obtaining ED values of about 500 Gy. But it is still not known whether this ED is close to the true ED: For example, Wieser et al. (1985) suggested a Mn normalization and recalculated a raw value of 610 Gy to an ED of 300 Gy, Although a certain decrease in the Mn 2÷ signals with additional doses has been observed (Hennig et al., 1981). Moreover, all three TL dates for this sample are in the ED range of 150-250 Gy. Isothermal annealing experiments carried out on this sample yielded a mean life of 7 × 106yr for the electron trap corresponding to the "dating" signal at g = 2.0007, whereas the signal at g = 2.0023 is related to a shallow trap level with a mean life of only 2 yr at 10°C (Fig. 6). Sample PARG

This sample is from a speleothem of unknown age in a cave in a quarry wall near Parnassos, Greece. The

5ample "'Hegy" 9 655 GHz 2row

Y

~ 5 G~p

2 Gp. *1.0krad

l 31.40

31.50

31.60 [G] Magnetic Field

FIG. 5. ESR spectra of the etched sample HEGY: The unstable peak at g = 2.0023. being absent in the natural sample, is detected by a narrow Gauss scan (0.5 Gpp, center), but is superimposed by a wider Gauss scan (2 Go0. below).

THE F I R S T I N T E R - L A B O R A T O R Y ESR C O M P A R I S O N PROJECT 220 •~ 10~ cu

1BO . .

160 120 100 80 . . . . . . .

.

60 .

temperature [ % ] ~,0 20 10 . . / '10? ~_~ /

Sample "HEOY"

// /

10~2,

m 106 :-:

/

g

105

/

/ /

.10=

/ "~= 2.0006

.103

/

10~ /

.102

/ /

109. / 10a.

/

/

/

/

/

/

107•

/

/

/

//g= 20023 lOS"

/

/

/

10~,

/

10 3. 10 ~

ooo2

o.obzs

o.oo3

o.o63s

951

erroneously high ED values (Skinner, 1983; Hennig and Grfin, 1983). It is generally possible to eliminate such interfering signals by using a small modulation, e.g. = 0.5 Gausspp, in connection with a sufficiently long scan time. In addition, interference by highly instable defects is minimized by a time gap of a few days between irradiation and ESR measurement. 6. The microwave energy applied to a calcite sample should be low: about 2 mW or less (Hennig and Griin, 1983). For quartz, energies as low as 0.05 mW are recommended by Yokoyama et al. 0985). 7. At least 6 aliquots should bc used for setting up growth curves and a strict statistical evaluation of the measured data (ESR signal, radiation dose) is essential. 8. Contamination by detritai components (e.g. clay) should be investigated and if necessary taken into account (Regulla et aL, 1985; Wieser et al., 1985). 9. The data should be presented or published in accordance with the recommendations given in the "Consensus of a Joint Meeting of ESR Groups" held during the fourth specialist seminar on TL and ESR dating (Schwarcz and Hennig, 1985).

1/T [K"~

Fla. 6. Arrhenius plot of the "'dating" signal (g = 2.0007) and the unstable signal at g =2.0023 for isothermal annealing experiments.

Sample PARG 9743 MHz 2row

o.sG~ corresponding to this signal in various calcite samples (Hennig and Grfin, 1983), but lower mean lives are a possiblity for some materials. 2. Another suitable signal is sometimes present at g = 2.0035. Because the corresponding trap has a lower, variable mean life (Table I) and frequently is sensitive to light, each sample must be checked for thermal stability and bleaching effects when using this g value. 3. If other signals are selected for dating purposes, it should be checked to see that they are absent in modern samples and whether anomalous fading occurs. The mean life of the trap must also'be evaluated. 4. Surface defects caused by grinding or milling usually appear at g = 2.0002. They can be minimized by removing at least 10% of the sample with dilute acid, e.g. HC] or acetic acid. 5. Other sources of interference may be unstable defects, e.g. the one at g = 2.0023. Prcannealing to empty such unstable traps is not recommended due to complex electron transfer processes which produce

not.

.~0002

S 3450

31,70

mogne tic field[G]

FIG. 7. ESR spectra of the natural and irradiated sample PARG.

952

G.J.

H E N N I G et al.

CONCLUSIONS The first E S R inter-laboratory comparison test shows that the E D data from calcites have a scatter of more than + 25% up to + 160% and do not yet fit the cited possible precision of + 5% (Hennig and griin, 1983). This is partially due to improper recording and evaluation of the spectra. However, further unknown, systematic errors must be found before the precision of E D evaluation of calcites will become better than + 25%. Acknowledgements--The Federal Ministry for Research

and Technology (BMFT) supported this study within the framework of a project on paleoelimate research. Special thanks are due the 16 laboratories listed below, who participated in this project. Our thanks are also due to Dr. Jfirgens of the Z~lpich office of the Rlaeinisehes Landesmuseum for supplying sample TROM, to Dr. Lanser of the Ruhrland Museum in Essen for supplying the bone sample, and to Dr. P. de Canniere of the University of Louvain-la-Neuve in Belgium for providing several spectra for this paper.

REFERENCES Aitken M. J. (1974) Physics and Archaeology, pp. 1-291. Clarendon Press, Oxford. Hennig G. J. and Griin R. (1983) ESR dating in Quarternary geology. Q. Sci. Rev. 2, 157-238. Hennig G. J., Herr W. and Weber E. (1981) ESR dating of speleothem of the Caune de l'Arago at Tautavel. Notes on problems and progress. In Datations. Absolutes et Analyses Isotopiques en Prehistoire--Methodes et Lirnites (Edited by de Lumley H. and Labeyrie J.)

pp. 551-556. CNRS, Paris. Hennig G. J., Griin R. and Brunnacker K. (1983) Interlaboratory Comparison Project of ESR-Dating, Phase I. PACT J. 9, 447~,52. Regulla D., Wieser A. and G6ksu H. Y. (1985) Effect of sample preparation on the ESR spectra of calcite, bone and volcanic material. Nucl. Tracks 10, 825-830. Schwarcz H, P. and Hennig G. J. (1985) Consensus of a joint meeting of the ESR groups. Nucl. Tracks (to be published). Skinner A. F. (1983) Overestimate of stalagmitic calcite ESR dates due to laboratory heating. Nature 3114, 152-154. Smith B. W., Smart P. L. and Symons M. C. R. (1985) ESR signals in a variety of speleothems calcites and their suitability for dating NucL Tracks 10, 837-844.

Wieser A., G6ksu H. Y. and Regulla D. (1985) Characteristics of gamma induced ESR spectra in various calcites. Nucl, Tracks 10, 831-836. Yokoyama Y., Falgueres C. and Qua6gebeur J. P. (1985) ESR dating of quartz from Quarternary sediments: first attempt. Nucl. Tracks I0, 921-928.

PARTICIPANTS Cowan D. L. and Rowiett R. M., Department of Physics, University of Missouri, Columbia, Missouri 65211, U.S.A. de Canniere P., Debuyst R. and Apers D., Laboratoire Inorganique et Nucleaire, Universit6 de Louvain, Chemin de Cyclotron 2, B-1348 Louvain-La-Neuve, Belgium. Dreybrodt W., Universit/it Bremen, Fachbereich Physik, bibliothekstrasse, D-2800 Bremen, F.R.G. G6ksu-Ogelmann Y., Cukurova University, Basic Science Faculty, Adana, Turkey. (Present address: c/o D. Regulla, see below). Griin R., Geologisches Institut der Universit/it zu K61n, Ziilpicher Strasse 49, D-5000 K61n 1, F.R.G. Hennig G. J., Nieders/ichsisches Landesamt fiir Bodenforschung, Stilleweg 2, D-3000 Hannover 51, F.R.G. Hiitt G., Institute of Geology of the Academy of Sciences of the Estonian S.S.R., Estonia Puistce 7, Tallinn 2000101, Estonian S.S.R., U.S.S.R. Massot J. C. and Valladas H., Institut de Chimie des Substances Naturelle, Av. de La Terrasse, CNRS, F-91190 Gif-sur-Yvette, France. Nambi K. S. V., Health Physics Division, Bhabha Atomic Research Center, Bombay-400 085, India. Pomeroy P. J. and Hunter D. S., University of Queensland, St. Lucia, Department of Chemistry, Brisbane, Queensland, Australia 4067. Radtke U., Geographisches Institut de Universit~it Diisseldorf, Universit~itsstrasse I, D-4000 Diisseldorf, F.R.G. Regulla D. and Wieser A., GeseUschaft fiir Strahlen- und Umweltforschung mbH Miinchen, Institut fiir Strahleuschutz, Ingolst~idter Landstasse 1, D-8042 Neuherberg, F.R.G. Sato T., Osaka University, Department of Physics, Faculty of Engineering Science, Toyonaka, Osaka, Japan. Singhvi A. K., Physical Research Laboratory, Navrangpura, Ahmedabad-380 009 India. Skinner A., Chemistry Department, Williams College, Williamstown, Massachussetts MA 01267, U.S.A. Smart P. L., Smith B. W. and Symons M. C. R., University of Bristol, Bristol BS8 ISS, England, U.K., and (M.C.R.S.), Department of Chemistry, University of Leicester, Leicester LEI 7RH, England, U.K.