Laboratrio Central de Metrologia

!PQ-LCN-LvO Junho 2005

REFERENCES
1. ISO 3507 - Laboratory glassware - Pyknometers, Genve 1999;
2. ISO 4787 - Laboratory glassware - Volumetric glassware - Methods for use and testing of capacity;
Genve, 1984;
3. M.G.Cox, The evaluation of key comparison data, Metrologia, 2002, Vol. 39, 589-595.
ABSTRACT
The EUROMET comparison Volume Calibration of a 100 ml Gay-Lussac Pycnometer, between fourteen
National Metrology Institutes (NMIs), was performed with the gravimetric method procedure. This paper
describes the volume instrument, the method and formula for volume determination, the calibration
procedure, the experimental conditions and the measurement results with the associated uncertainties.
This comparison was expected to be representative for all type of laboratory glassware.
DATA EVALUATION AND UNCERTAINTY ANALYSIS IN AN INTERLABORATORY COMPARISON OF A
PYCNOMETER VOLUME
ELSA BATISTA, EDUARDA FILIPE
Portuguese Institute for Quality (IPQ)
Rua Antnio Gio, 2 Caparica - Portugal
STATISTICAL DATA
The determination of the statistical parameters: reference value, uncertainty of the
reference value and a chi-square test were performed according to M.G.Cox [3].
Consistency statistical test - Chi-square test
To identify the inconsistent results a chi-square test can be applied to all results.
degrees of freedom: = N -1
Regard the consistency check as failing if:
The obtained weighted mean using the 14 laboratories is: y = 100,0917 ml with a
u(y) = 0,0006 ml for k=2
The chi-square test gives the following results: ;
The consistency test fails. The value for one of the laboratories SLM:
>
The volume value for the SLM is then removed from the weighted mean calculation and a
new consistent test is performed. The results now are the following:
;
In conclusion the results are consistent and a new value for the weighted mean was
obtained: x
ref
= 100,0914 ml, u(x
ref
) = 0,0006 ml with k=2.
) ( / 1 .... ) ( / 1
) ( ... ) (
2
1
2
2
1
2
1
n
n n
x u x u
x u x x u x
y
+ +
+ +
=
) ( / 1 ... ) ( / 1
1
) (
2
1
2
n
x u x u
y u
+ +
=
) (
) (
...
) (
) (
2
2
1
2
2
1 2
n
n
obs
x u
y x
x u
y x
+ +

=
{ } 05 , 0 ) ( Pr
2 2
< >
obs

3620 , 22 ) 13 ; 05 , 0 (
2
= 3602 , 41
2
=
obs

4622 , 26
2
= SLM 3620 , 22 ) 13 ; 05 , 0 (
2
=
0261 , 21 ) 12 ; 05 , 0 (
2
= 3879 , 14
2
=
obs

Determination of the reference

value
Determination of the reference
value uncertainty
Figure 3 Laboratory results with reference value
Figure 4 Mean average of presented uncertianties
INTRODUCTION
Volume measurement is critical in many laboratories and industries. In order to identify
and reduce possible errors in intensive liquid handling process it is necessary to calibrate
the used volumetric glassware equipment.
The majority of the National Metrology Institutes (NMIs) performs this type of calibration.
With the purpose of comparing the experimental calibration procedure and uncertainties a
EUROMET 692 comparison was performed with 14 participants NMIs. The calibration of
Gay-Lussac pycnometer of 100 ml, representative for all type of laboratory glassware, was
carried on and IPQ, as the pilot laboratory, provided the pycnometer volume standard
(figure 1).
The main reasons for choosing this type of pycnometer were:
Easy handling;
The volume cannot be changed unless breaking the instrument;
Easy cleaning;
Possibility to observe air bubbles.
The Instrument
A pycnometer is an instrument used for
the measurement of the density of
different liquids from water to high
viscosity paints. For a correct density
measurement, the volume of the
pycnometer must be obtained by
calibration using a gravimetric method.
Figure 1 Gay-Lussac pycnometer of 100 ml
The suggest method to perform the
pycnometer calibration was the gravimetric
method, were the calibration liquid is
distilled water at a reference temperature
of 20 C. The following formula described
in ISO 4787 can be used for the calculation
of the contain volume:
CONCLUDING REMARKS
This comparison involved 14 laboratories and lasted one and a half year. One of the major
risks was to break the glass pycnometer and this occurred after 5 measurements.
Replacing the pycnometer and adding a correction to the first 5 volume results solved the
problem.
Globally the results are quite satisfactory, the maximum and minimum reported volumes
differ less than 0,01%. With the exception of two participants, the laboratories volume
results were quite consistent with the reference value and with each other. The uncertainty
budgets are very similar and the major uncertainty component to the final uncertainty was,
for the majority of the participants, the water density.
UNCERTAINTIES
100,0760
100,0810
100,0860
100,0910
100,0960
100,1010
100,1060
100,1110
100,1160
!P
Q
C
N
!
B
N
N
-L
N
E
F
O
R
C
E
P
T
B
N
N
i-v
S
L

S
L
N
U
N
E
C
E
N
!N
G
C
O
N
H
E
!N
B
E
v
S
P
Laboratories
V
o
l
u
m
e

{
m
l
)
volume
Reference value
Expanded uncertainty of the reference value
+f- 0,01
Average of the standard uncertainties components
0,00000
0,00020
0,000+0
0,00060
0,00080
0,00100
0,00120
0,001+0
0,00160
N
a
ss p
ie
c
e
s d
e
n
sity
W
a
te
r d
e
n
sity
A
ir d
e
n
sity
E
x
p
a
n
sio
n
co
e
fic
ie
n
t
T
e
m
p
e
ra
tu
re
N
a
ss fille
d
O
th
e
r
S
ta
n
d
a
rd
d
e
v
ia
tio
n
Standarduncertainties
V
o
l
u
m
e

{
m
l
)
Components taken in account: mass
determination, mass pieces density,
air density, water density, expanded
coefficient, temperature and other
smaller and specific components.
As can be seen the major source of
uncertainty is the water density
followed by the mass determination.
Figure 3 Laboratories results compared with
reference value
THE METHOD
Correction of the measured results
During the comparison it was necessary to use two Gay-Lussac pycnometers because the
first one (with the number 62) was broken after the participation of five laboratories and
replaced by a new one with number 144.
To compare the two groups of results a correction was applied to the results of the first
pycnometer.
The correction was obtained averaging the difference of the values obtained by the
laboratories that performed the calibration of both pycnometers, PTB and IPQ.
Table 1- Correction of the volume
The value 0,1118 ml was then added to the determined volume of the five laboratories that
performed the calibration of the pycnometer n 62.
[ ] ) 20 ( 1 1
1
) (
20

= t I I V
B
A
A W
E L

Figure 2 Weighing scale and F2 mass

Difference (ml) Average
(ml)
Uncertainty
(ml)
IPQ
0,1109
IPQ1 - IPQ2
0,1118 0,0016
PTB
0,1120
PTB1 - PTB2


The suggest method to perform the
pycnometer calibration was the gravimetric
method, were the calibration liquid is
distilled water at a reference temperature
of 20 C. The following formula described
in ISO 4787 can be used for the calculation
of the contain volume: