1. SP Sveriges Provnings- och Forskningsinstitut Euromet project 691 Calibration Inter- comparison of a 5-litre volume glass standard P. Lau

2. SP Sveriges Provnings- och Forskningsinstitut Task: ”Perform a volume calibration!” How many laboratories? 20 from 19 countries Which volume?”dry, contained” – below the mark ”wet, delivered or poured” How many measurements? contained 18 labs:4 (1 lab) 8 (1 lab) 10 (14 lab) 11 (1 lab) 15 (1 lab) delivered 16 labs:15 (3 lab) 10 (13 lab) 15 laboratories performed both calibrations What about cleaning? no instructions given Pouring times? according to laboratory practice How to report uncertainty? form but no instructions given

3. SP Sveriges Provnings- och Forskningsinstitut Result ”contained” or ”dry” volume mean median  0,01 %  0,13 ml  1,1 ml 11841510 measurements  2 s

4. SP Sveriges Provnings- och Forskningsinstitut Result “delivered”, “wet” or ”poured” volume mean median  0,01 %  0,16 ml  1,1 ml 15 10 measurements  2 s

5. SP Sveriges Provnings- och Forskningsinstitut 00,20,40,60,81,01,2 CEM EIM SMU CMI SP BEV FORCE SMD UME NWML NMi METAS SLM IMGC NMI OMH IPQ BNM-LNE PTB GUM 00,20,40,60,81,0 CEM EIM SMU CMI SP BEV FORCE SMD UME NWML NMi METAS SLM IMGC NCM OMH IPQ BNM-LNE PTB GUM Uncertainty (k=2) – range and standard deviation Uncertainty Range Stdev Uncertainty Range Stdev Poland Germany France Portugal Hungary Bulgaria Italy Slovak Rep. Switzerland Netherlands United Kingdom Turkey Belgium Denmark Austria Sweden Czech Rep. Slovak Rep. Greece Spain ”contained” volume”delivered” volume ml

6. SP Sveriges Provnings- och Forskningsinstitut Distribution of uncertainty components weighing 1 weighing 2stdevwater density meniscuswater densityweighing 1weighing 2 stdevwater densityair densityweighing meniscusstdevweighingwater density meniscusstdevwater densityweighing meniscuswater densitystdevweighing 1,10 0,70 0,53 0,46 0,45 0,42 0,38 0,37 0,35 0,31 * 0,29 0,26 0,22 0,16 * 0,14 0,13 meniscustemperaturestdevwater density stdevweighingwater densitytemperature meniscustemperaturewater densitystdev meniscuswater densityweighing stdevwater densitytemperatureweighing stdevweighingwater densitymeniscus stdevwater densityair densityweighing stdevmeniscuswater densityweighing meniscusstdevweighingtemperature stdevwater densityair densityweighing meniscusstdevwater densityweighing water densitystdavmeniscus weighing stdevtemperatureweighing 1weighing 2 komp. 1 komp. 2 komp. 3 komp. 4 1-sigma level * values for GUM & PTB refer to delivered volume U(contained volume)

7. SP Sveriges Provnings- och Forskningsinstitut Cross-correlation – “degree of equivalence” CMIBNM-LNESMUSLMPTBGUM ±2 s area

8. SP Sveriges Provnings- och Forskningsinstitut Uncertainty declaration Just one figure – Force (1) 5 components – SMD, NWML (2) 6 components – Metas, UME (2) 7 components – IPQ, BNM-LNE, SMU, NMi, BEV (5) 8 components – CMI, SLM, CEM, OMH (4) 9 components – SP, GUM, NCM (3) 11 components – EIM (1) 14 components – PTB (1)

9. SP Sveriges Provnings- och Forskningsinstitut Possible conclusions from the  The total outcome is satisfactory.  One lab out of 20 (5 %) (in 35 results) just outside 95 % confidence level.  Large difference in repeatability between the laboratories (is low spread always good?).  Large difference in uncertainty declarations.  Some results (at least 5 for dry and 6 for wet volume) are unrealistic small.  Probably 3 to 5 components really matter.  A good and realistic uncertainty estimation depends more on a ”feeling” than on a sophisticated modelling.  In the perspective of an inter-comparison systematic effects (connected with a method) are serious uncertainty contributions (cleaning, pouring, experimental realization).  The meniscus setting is very personal (random within laboratory – systematic in global aspect).  In my opinion uncertainty declarations need to incorporate the limitations of a calibration object.