CONTRIBUTED BY: THOMAS HIGHAM

Much of the information presented in this section is based upon the Stuiver and Polach (1977) paper "Discussion: Reporting of C14 data". A copy of this paper may be found in the Radiocarbon Home Page

The radiocarbon age of a sample is obtained by measurement of the residual radioactivity. This is calculated through careful measurement of the residual activity (per gram C) remaining in a sample whose age is Unknown, compared with the activity present in Modern and Background samples. You can get an idea of the relationship between C14 and age at the Carbon Dating calculator page.

Modern standard

The principal modern radiocarbon standard is N.I.S.T (National Institute of Standards and Technology; Gaithersburg, Maryland, USA) Oxalic Acid I (C2H2O4). Oxalic acid I is N.I.S.T designation SRM 4990 B and is termed HOx1. This is the International Radiocarbon Dating Standard. Ninety-five percent of the activity of Oxalic Acid from the year 1950 is equal to the measured activity of the absolute radiocarbon standard which is 1890 wood. 1890 wood was chosen as the radiocarbon standard because it was growing prior to the fossil fuel effects of the industrial revolution. The activity of 1890 wood is corrected for radioactive decay to 1950. Thus 1950, is year 0 BP by convention in radiocarbon dating and is deemed to be the 'present'. 1950 was chosen for no particular reason other than to honour the publication of the first radiocarbon dates calculated in December 1949 (Taylor, 1987:97).

The Oxalic acid standard was made from a crop of 1955 sugar beet. There were 1000 lbs made. The isotopic ratio of HOx I is -19.3 per mille with respect to (wrt) the PBD standard belemnite (Mann, 1983). The Oxalic acid standard which was developed is no longer commercially available. Another standard, Oxalic Acid II was prepared when stocks of HOx 1 began to dwindle. The Oxalic acid II standard (HOx 2; N.I.S.T designation SRM 4990 C) was made from a crop of 1977 French beet molasses. In the early 1980's, a group of 12 laboratories measured the ratios of the two standards. The ratio of the activity of Oxalic acid II to 1 is 1.29330.001 (the weighted mean) (Mann, 1983). The isotopic ratio of HOx II is -17.8 per mille. There are other secondary radiocarbon standards, the most common is ANU (Australian National University) sucrose. The ratio of the activity of sucrose with 0.95 Ox was first measured by Polach at 1.50070.0052 (Polach, 1976b:122). Later inter-laboratory measurements put the ratio at 1.5081 (Currie and Polach, 1980).

According to Stuiver and Polach (1977), all laboratories should report their results either directly related to NBS Oxalic acid or indirectly using a sub-standard which is related to it.

Background

It is vital for a radiocarbon laboratory to know the contribution to routine sample activity of non-sample radioactivity. Obviously, this activity is additional and must be removed from calculations. In order to make allowances for background counts and to evaluate the limits of detection, materials which radiocarbon specialists can be fairly sure contain no activity are measured under identical counting conditions as normal samples. Background samples usually consist of geological samples of infinite age such as coal, lignite, limestone, ancient carbonate, athracite, marble or swamp wood. By measuring the activity of a background sample, the normal radioactivity present while a sample of unknown age is being measured can be accounted for and deducted.

In an earlier section we mentioned that the limit of the technique is about 55-60 000 years. Obviously, the limit of the method differs between laboratories dependent upon the extent to which background levels of radioactivity can be reduced. Amongst accelerator laboratories there has been mooted the theoretical possibility of extended range dating to 75 000 yr +, at present this seems difficult to attain because of the problems in accurately differentiating between ions that mimic the mass and charge characteristics of the C14 atom. Beukens (1994) for instance has stated that this means the limit of the range for his Isotrace laboratory is 60 000 yr which is very similar to the conventional range.



Figure 1: This gif shows the comparison in radioactivity between a sample, or unknown (green area) , a modern standard (dark blue) and a background (small red peaks) derived from beta decay. The scale represents log E (energy).

Conventional radiocarbon ages (BP)

A radiocarbon measurement, termed a conventional radiocarbon age (or CRA) is obtained using a set of parameters outlined by Stuiver and Polach (1977), in the journal Radiocarbon. A time-independent level of C14 activity for the past is assumed in the measurement of a CRA. The activity of this hypothetical level of C14 activity is equal to the activity of the absolute international radiocarbon standard.

The Conventional Radiocarbon Age BP is calculated using the radiocarbon decay equation:

t=-8033 ln(Asn/Aon)

Where -8033 represents the mean lifetime of 14C (Stuiver and Polach, 1977). Aon is the activity in counts per minute of the modern standard, Asn is the equivalent cpm for the sample. 'ln' represents the natural logarithm.

A CRA embraces the following recommended conventions:

Three further terms are sometimes given with reported radiocarbon dates. d14C, D14C and deltaC13.

All are expressed in per mille notation rather than per cent notation (%).
d14C represents the per mille depletion in sample carbon 14 prior to isotopic fractionation correction and is measured by:

d14C=((Asn/Aon) - 1)1000 per mille

D14C represents the 'normalized' value of d14C. 'Normalized' means that the activity is scaled in relation to fractionation of the sample, or its deltaC13 value. All D14C values are normalized to the base value of -25.0 per mille with respect to the standard carbonate (VPDB). D14C is calculated using:

D14C=d14C - 2(dC13 + 25)(1 + d14C/1000) per mille

This value can then be used to calculate the CRA using the equation given above.

Radiocarbon age=-8033 ln(1 + D14C/1000)



Figure 1: Decay curve for C14 showing the activity at one half-life (t/2). The terms "%Modern", or "pmC" and D14C are shown related in this diagram along with the Radiocarbon age in years BP (Before 1950 AD).

Age reporting

If the reservoir corrected conventional radiocarbon age calculated is within the past 200 years, it should by convention be termed 'Modern' (Stuiver and Polach, 1977:362). If a sample age falls after 1950, it is termed greater than Modern, or >Modern. Absolute percent modern (%M or pMC - 'percent modern carbon') is calculated using:

%M=100 x Asn/Aabs

or,

Asn/Aon(1/8267(y-1950)) x 100 percent

Where Aabs is the absolute international standard activity, 1/8267 is the lifetime based on the new half life (5730 yr), Y = the year of measurement of the appropriate standard. This is an expression of the ratio of the net modern activity against the residual normalised activity of the sample, expressed as a percentage and it represents the proportion of radiocarbon atoms in the sample compared to that present in the year 1950 AD. Thus, %Modern becomes a useful term in describing radiocarbon measurements for the past 45 years when, due to the influx of artificial radiocarbon into the atmosphere as a result of nuclear bomb testing the 'age' calculation becomes a 'future' calculation.

If the sample approaches D14C = -1000 per mille within 2 standard deviations, it is considered to be indistinguishable from the laboratory background, ie, not able to be separated with confidence from the laboratory countrates which result from a sample which contains no radionuclide. In this instance, a minimum age is calculated. An example of a minimum age is >55, 000 yr or >50, 000 yr (Gupta and Polach, 1985).

Should the activity of the sample be indistinguishable from the background activity at 1 standard deviation, it is released as background.

Samples whose age falls between modern and background and are given finite ages.Standard errors released with each radiocarbon assay (see below) are usually rounded by convention (Stuiver and Polach, 1977). Again, not all laboratories subscibe to these conventions, some do not round up ages.

Age (yr) Radiocarbon date rounded to Error ( value)
0 - 1000 nearest 10 nearest 5 up
1000 - 10000 nearest 10 nearest 10 up
10000 - 25000 nearest 50 nearest 10 up
>25000 nearest 100 nearest 50 up


Standard error

We mentioned previously that statistical analysis is necessary in radiocarbon dating because the decay of C14 although constant, is spontaneous. It is not possible to measure all of the radioactivity in a given sample, hence the need for some kind of statistical analysis of counted data. The distribution of counted C14 decay events will, over time, yield a pattern. The pattern is termed a "normal distribution curve". A normal or "Gaussian" distribution describes the symmetrical bell shaped cluster of events around the average or mean of the data. In a normal distribution, 2 out of 3, or 68% of the values or counts observed will fall within one standard deviation of the average of the data. At two standard deviations, 95% of the observed counts will fall within the range and at three standard deviations, 99% of the counts which comprise the normal distribution will fall within this region. Each radiocarbon date is released as a conventional radiocarbon age with 'standard error'. This is the '' value and by convention is 1 sigma. The standard error is based principally upon counting statistics (but see below).

According to Stuiver and Pearson (1992), the majority of laboratories report the measured counting statistics as a laboratory standard error. This results in considerable under-reporting (Scott, Long and Kra, 1990:253-393; Stuiver and Pearson, 1992:20) because a standard error based upon counting statistics alone does not include additional errors which may be incurred, for example, during sample preparation. Pearson (1979:21) has suggested:

"Those involved in radiocarbon dating should be alert to the various possible sources of error and recognize that the precision quoted on a date may be quite unrealistic if the error sources have not been investigated in detail".
Gordon Pearson. 1979. "Precise 14C Measurement by LS Counting". Radiocarbon 21(1):1-22.

To present a date with a realistic standard error, Pearson (1980; 1983) and Pearson et al. (1986:929) have identified the factors which contributed significantly to errors in beta counting using Liquid Scintillation spectrometry. Ostensibly, this research was a precursor to high precision dating of Irish bog oak samples for the calibration of the radiocarbon timescale (Pearson, 1980, 1983; Pearson and Stuiver, 1993). He investigated each principal factor contributing to errors and considered their effect on overall laboratory precision and concluded that a standard error of 25 radiocarbon years was possible in the Belfast laboratory. Standard errors quoted by the Belfast laboratory, then, are based upon this analysis.

Many laboratories today calculate a laboratory error multiplier to account for all errors account for routine variation in reproducibility in radiocarbon dating. Stuiver (Stuiver and Pearson, 1993), for instance, has reported that the standard errors reported in the University of Washington laboratory results are based a lab multiplier. According to Stuiver and Pearson (1993), the error multiplier (or 'K') is a measure of the laboratory reproducibility, incorporating the errors resulting from the preparation of gas, its loading, memory effects and counting statistics. 'K' is defined as the actual standard error divided by the quoted standard error and is usually generated through repeat dating of a standard of known age or consensus age.

In addition, inter-laboratory comparison exercises have been undertaken to evaluate laboratory variation and identify causes for it. We hope to include some information regarding these important programmes run principally by the International Atomic Energy Agency (IAEA) and Glasgow University Department of Statistics.


Accuracy and Precision in Radiocarbon dating

It is important to note the meaning of "accuracy" and "precision" in radiocarbon dating. Accuracy refers to the date being a 'true' estimate of the age of a sample within the range of the statistical limits or value of the date. Thus, for the sake of argument, if we were radiocarbon dating a sample of Harold 1's (d. 1066) remains, and obtained a date of 104040 AD, we would have dated the event of his death accurately. If however the date were 100015 AD, we would be inaccurate. In terms of precision, however, the former is imprecise in comparison to the latter. It can be seen that the date of 100015 AD while being highly precise is, in this instance, inaccurate.

 INDEX | Introduction | Measurement | Applications | WWW Links | k12 | Publication | Corrections |
Age calculation | Calibration | Pretreatment | References | Awards | Credits | What's new? | Email |

HTML DOCUMENT BY T.HIGHAM.