Radiocarbon dating is universally used as an essential dating tool in the archaeological and earth (Quaternary) sciences. The technique has enjoyed considerable success with ongoing developments in both the sophistication of experimental practice and an ever-widening range of applications. Most recently (since the 1980s), a new generation of laboratories has been created, based on the exploitation of accelerator mass spectrometry (AMS) for the differentiation and measurement of carbon isotope abundances in natural materials. Worldwide, there are over 100 14C laboratories now operational in universities, research organizations, museums, and as commercial enterprises. There is an inevitable diversity of experimental approaches and applied priorities within these facilities. Some are well established, while others are relatively recent members of the international 14C community. Consequently, as a group, the laboratories reflect to varying extents the progress achieved over several decades of experience and methodological options. Furthermore, since progress in archaeology and related earth sciences cannot respect geographical and/or present political boundaries, there has been, and continues to be, an inevitable consequence that sample materials from specific cultural contexts are submitted to different laboratories and at different times. In this situation, the issue of comparability between results and amongst laboratories becomes paramount. Users of the results from 14C dating are also concerned with the comparability and quality of laboratory results and the quality assurance programs that laboratories undertake are thus important in ensuring user confidence. The harmonization of measurements and the traceability of results to internationally recognized standards are also major goals of the program of work described in this special issue.

The sets of core samples were distributed to over 120 laboratories that had returned an original questionnaire seeking expressions of interest in participation. A reporting format for the results was also agreed and distributed to the laboratories at the same time. This is shown in Table 2.1. Laboratories were originally given 1 yr (i.e., to August 2000) to complete the analyses and return the results, but this was later extended to December 2000. In this section, we briefly describe the laboratory characteristics and the overall response rate of the participating laboratories.

In this section, we present the exploratory analysis of the results submitted by the extended deadline of December 2000. We first deal with Samples C.J, before considering the near-background samples A and B (Kauri wood). The aims of the exploratory analysis are to discover the range of results reported for each sample and the initial evaluation of the effects of any factors that might be a source of variation in the results. For each sample, in turn, we consider the main summary statistics.the number of results reported (N), their mean or average, median, the standard deviation (StDev), the standard error of the mean (Sem), the quartiles (25th [Q1] and 75th [Q3] percentiles), and the minimum (Min) and maximum (Max)—before graphically studying the overall distribution of results in the form of a boxplot, with a view to identifying any extreme or outlying observations. The summary statistics and distribution of results for each laboratory type are also shown. Further details on the statistical methods used are contained in Appendix 3.

The design of FIRI is such that for each laboratory, we have some basic, though limited, information on the laboratory procedures, including the method of pretreatment applied to the samples, the modern standard, and the background material used. These can be considered as factors in the experiment and through statistical analysis, we can investigate whether they offer a statistically significant explanation of the observed variation. The different levels of the factors are described in Table 4.1. In addition, the laboratory type is also considered as a further factor (with 3 levels of LSC, GPC, and AMS).

The design of FIRI included 3 pairs of duplicate samples: A and B (Kauri wood) near background, D and F (Belfast wood) around 50 pMC, and G and J (barley mash) at 111 pMC. Why include duplicates? Duplicates by their nature allow us to explore the within-lab variability and to assess whether the quoted errors are representative. We can also explore the differences as a function of the sample activity. In this section, we explore the differences between the duplicates. We also consider some different graphical presentations. First, we summarize the differences, then graphically explore the boxplot (to consider the distribution of differences), then a scatterplot of the duplicate pair (to show correlation and reproducibility), and finally, a measure of agreement plot (Bland and Altman 1999). The horizontal axis in this final plot is the mean of the duplicate pair and the vertical axis is the difference in the duplicate pair. Agreement between the pairs would result in the points being randomly scattered around the horizontal zero line.

Kauri wood, a sub-fossil wood from New Zealand (which had previously been used in an IAEA exercise, IAEA-C4, in 1990), was considered to be an important sample to include in FIRI because it provided a link to previous exercises, was available in sufficient quantity, and was a “close to background” organic sample. IAEA-C4 had previously been criticized since it was believed that in its milling, some contamination had been introduced, so that a replacement sample would prove useful. The Kauri wood has a very low 14C activity and, as such, is very sensitive to even small amounts of contaminant carbon. Such low-activity samples give a true test for the laboratory procedures since pretreatment and laboratory background definition become critical.

Each material needs to be characterized by estimating its activity, which creates a reference value for each material. This value then can be considered as the “known” activity of the material and future analyses can be compared to this to quantify the accuracy of the measurement. In this way, the materials remain useful for laboratory quality assurance.

As part of the FIRI program, it was recognized that providing samples in sufficient quantity for laboratory procedures is close to ideal and does not represent “typical” conditions. Therefore, laboratories were asked to consider 2 optional studies: investigating the effects of sample size on results and achieving high precision. The sample size study was focused on Sample E, humic acid, which had been chosen because of the rigorous pretreatment it had undergone in the solution stage, and which would ensure sample homogeneity. The precision study was focused on Sample D, the Belfast dendro-dated wood sample, given its importance in the master calibration.

After the first and main phase of FIRI, which focused on routinely measured materials, an optional series of samples were also made available to participating laboratories. This second list included archaeological samples, mammoth tusks, and modern cellulose. Not all samples were available in sufficient quantity for radiometric measurement (in particular, the mammoth tusks). The samples are briefly described in Table 9.1 below.

A significant and substantial effort has been made by the 14C community in quality assurance (QA) procedures, of which participation in FIRI is only one part but one that provides an independent and blind check on laboratory performance. The overwhelming willingness to participate is a testament to the importance which laboratories place on quality.

TIRI was officially launched at the 14th International Radiocarbon Conference in Arizona in 1991. Prior to the conference, 150 laboratories received a letter describing the general intention to organize an intercomparison and over 90 laboratories from around the world responded positively to the invitation to participate. Simply stated, the aims of this intercomparison were:

1. 1. To function as the third arm of the quality assurance (QA) procedure.

2. 2. To provide an objective measure of the maintenance and improvement in analytical quality.

3. 3. To assist in the development of a “self-help” scheme for participating laboratories.