| As recommended at the meeting of CO2 experts at Lake Arrowhead, October 1990, an inter-laboratory intercomparison of standard gases was conducted during 1991-1992. Dr P. Tans of NOAA/USA served as the coordinator and Dr J. Perterson served as referee. |
| Two sets of three tanks were provided by the US National Institute of Standards and Technology (NIST). The NIST analysed each tank at the beginning and end of each circuit. The second and next-to-last analysis of each tank was done at the US National Oceanic and Atmospheric Administration (NOAA). These "before-and-after" analyses by both NIST and NOAA indicated that the CO2 concentrations in the tanks did not change during the period of the intercomparison. |
| One set of tanks - set A - was circulated among Laboratories in Europe. These were located in the Federal Republic of Germany, France, Spain (Canary Islands), Italy and Hungary. The other set of tanks - set B - was circulated among Laboratories in the USA (Scripps), Canada, Japan, New Zealand and Australia. |
| Each Laboratory was given only the approximate value of the CO2 concentration in each tank. They were instructed to treat each tank as containing a gas of unknown concentration. They were to analyse the gas using their normal laboratory procedures and report the finding to the referee. The referee kept all reports confidential until all reports were received. |
| The results shown in Table 1 are those initially reported directly to the referee by each participant. The Laboratories are listed in chronological order of analysis for each set of tanks. For those Laboratories who reported several determinations of each tank, the concentrations listed in the Table are the mean of the values they reported. The missing value for New Zealand is because the concentration of about 375 ppm for the high tank was beyond the range of their suite of standard tanks. |
| To intercompare the results of all Laboratories with each other, both set A and set B, see Table 2. Here, the reports of each Laboratory (from Table 1) are listed as the difference from the USA-NOAA value for that set. (The USA-NOAA value was subtracted from the Laboratory value). The USA-NOAA value used here is the mean of the values reported by them at the beginning and at the end of the intercomparison. Using the USA-NOAA value here as a reference does not imply that their numbers have absolute accuracy. |
| The results reported below should be considered as "preliminary" concentrations. Any drift of Laboratory standards and/or changes in the Scripps Primary Scale have not been accounted for. Moreover, some Laboratories had very recent ties to the CCL: scale whereas others did not. Consequently, some institutions have already stated that their final analyses will change significantly and the experts meeting in Rome agreed that all participants should have an opportunity to submit updated numbers. The first step in calculation of final values is for the Scripps CCL to provide to participants the most up-to-date values of any standard tanks calibrated by them over the last few years. This had net yet been completed as this report was finished. |
| As assessment of the preliminary results indicates that most, but certainly not all, Laboratories had fair agreement (see Table 2). At the low concentration (341 ppm), most Laboratories were within 0.4 ppm of each other, but at the high range (375 ppm) less than half agreed within 0.4 ppm. |
| Even though the inter-Laboratory differences will almost certainly narrow when final values are reported, marked improvement is needed to meet the experts' recommended international network precision of 0.05 ppm. |
