The impact of cleaning of weights before calibration


ACTA IMEKO | www.imeko.org December 2020 | Volume 9 | Number 5 | 7 

ACTA IMEKO 
ISSN: 2221-870X 
December 2020, Volume 9, Number 5, 7 - 11 

 
 

THE IMPACT OF CLEANING WEIGHTS BEFORE CALIBRATION 
 

Christian Müller-Schöll1 

 
1 Mettler-Toledo International Inc., Greifensee, Switzerland, christian.mueller-schoell@mt.com  

 

Abstract: 

This paper investigates the necessity of cleaning 

weights before calibration. The conflict between 

cleaning a weight and breaking the history by 

cleaning is analyzed. Experience from many 

hundreds of calibrations with measurements before 

and after cleaning is reported and analyzed. It is 

concluded, that weights should be cleaned in order 

to reach a stable state but should also be calibrated 

before and after cleaning to gain information about 

metrological traceability and thus the validity of the 

results that were produced with the weights. 

Keywords: calibration; weights; cleaning; 

contamination; OIML R111-1; traceability 

1. INTRODUCTION 

Cleaning of physical material measures for mass 

has been a topic since the making of the first 

kilogram standards in the 18th century.  

This paper details on the necessity and the effects 

of cleaning of industrial weights which are used for 

testing or calibration of weighing instruments. For 

this, a pool of some hundred calibrations of weights 

before and after cleaning are investigated. 

Section 2 sets the background for the historically 

documented need for cleaning weights. Section 3 

treats the question if cleaning requires calibrations 

before and after, while section 4 raises the question 

of safety that is gained by cleaning. Section 5 gives 

a practical example of a laboratory’s policy 

regarding cleaning of weights. In section 6 we detail 

the results of one specific weight, while in section 7 

we shed light from different angles on the question 

of significance. In section 8 we present results of 

some hundred weights’ calibrations and in section 9 

we investigate the effect of cleaning on metrological 

traceability. Section 10 gives an overview of the 

conclusions. 

2. BACKGROUND FOR CLEANING 
WEIGHTS 

From the beginning of modern mass metrology 

in the 18th century, metrologists have considered 

cleaning of mass artefacts: even the IPK, which 

served as the definition of the kilogram until 2019, 

assumed its exact mass only after a cleaning and 

washing procedure (French: “nettoyage-lavage”) 

which apparently removed as much as 15 µg from 

the artefacts at the BIPM [1]. 

Obviously, the artefacts reach a stable mass 

value only in a “clean” state whereas the “unclean” 

state is instable, unreliable and dependent on 

external influences like time since last cleaning, 

intensity, frequency, and kind of usage etc. 

Still today, in OIML R111-1, the step “cleaning 

weights” is listed as a mandatory element in the 

procedure for the assessment of conformity and 

OIML states “It is important to clean weights before 

any measurements are made…” [2]. However, no 

specific method is prescribed. The calibration or 

verification laboratory may choose appropriate 

methods according to their expertise, competence 

and capabilities. 

Although the definition of the kilogram has been 

freed from being dependent on an artefact in 2018, 

weight pieces are still an indispensable device for 

establishing a mass scale, for calibrating weights 

and for calibrating and testing weighing instruments. 

It is worth noting that immediately after cleaning 

the surfaces of weights are likely to be in an unstable 

state. Therefore, it is necessary to allow for a certain 

stabilisation period (B.4.2 in [2]) whose length is 

dependent on the cleaning method and the accuracy 

class of the weights. 

The reason for cleaning is obviously the fact that 

weight pieces are frequently contaminated and that 

the “clean state” is the only reliable and 

reproducible state, while the “dirty state” is rather 

volatile. In fact, the mass of the dirty state can be 

reduced as unintentionally (by randomly wiping 

away dirt) as it can be raised (by touching or 

contaminating the weight). With a reliable and 

reproducible and non-abrasive cleaning method, it 

should be possible to bring a piece of stainless steel 

always to the same mass (within measurable 

reproducibility). 

3. IS CLEANING AN INTERVENTION? 

ISO 17025 requires that calibration results shall 

be collected, “before and after any adjustment or 

repair, if available” [3]. The intention of this 

http://www.imeko.org/
mailto:christian.mueller-schoell@mt.com


ACTA IMEKO | www.imeko.org December 2020 | Volume 9 | Number 5 | 8 

requirement is obviously that an uninterrupted 

history of the device under calibration is available 

and the user of the object can use the information of 

the “as found” state in order to assess the 

appropriateness of handling, of assumed accuracy 

and stability and finally of calibration intervals. 

Unfortunately, [3] limits the applicable cases in an 

exhaustive list to “adjustment” and “repair”. It is 

clear that a cleaning intervention on a weight might 

have the same effect as an adjustment intervention 

and thus should be considered in the same way. 

The strict application of the ideas of [3] together 

with the procedure given in [2] must lead to the 

conclusion that all weights have to be cleaned 

before calibration and all calibrations have to 

consist of a set of two full calibrations, one before 

and one after the cleaning. The according 

calibration certificate is then a so-called “as-found–

as-left-certificate”. Not collecting data of the state 

before cleaning would – strictly speaking – result in 

a “break in the history” of the measuring instrument. 

4. SAFETY CONSIDERATIONS 

The fact that measurement instruments can carry 

over any kind of contamination from the 

environment they are used in, into the calibration 

laboratory, is frequently under-estimated, according 

to the view of the author. This applies for both 

hazardous contaminations as well as for “just dust 

and dirt” in the case of weight pieces. Both kinds of 

contaminations mentioned above endanger the 

integrity of the calibration laboratory equipment, 

and, even more important, endanger the integrity of 

un-involved third parties’ equipment, when 

contamination is carried from an instrument onto 

the laboratory equipment and from there to third 

party customers’ instruments. 

5. THE POLICY OF THE SWISS 
CALIBRATION LAB OF METTLER 

TOLEDO 

The Laboratory of Mettler-Toledo GmbH in 

Switzerland calibrates many thousands of weights 

per year. Customers expect calibration values of a 

stable state of their weights. Unfortunately, most 

customers are not willing to pay for the effort of a 

double calibration (before and after cleaning). 

As a compromise between the pure metrology of 

not changing instruments without additional 

calibration and the minimisation of the risk of 

contamination, and the request of low prices and 

stable and consistent calibration results, the policy 

generally applied for weight calibrations is as 

follows (and this policy is made known to the 

customer): 

• Weights will be cleaned generally to protect the 
laboratory and un-involved parties’ artefacts. 

• E1 class weights will not be cleaned. 

• If a customer wants a calibration of the un-
cleaned state, the risk for the laboratory is 

estimated. In severe cases, the bottom surface of 

the weight is cleaned in order to protect the 

comparator balances from dirt. The customer 

may choose if the weight shall be cleaned and 

calibrated again. 

6. EFFECT OF CLEANING WEIGHTS 

In contrast to most customers of the calibration 

laboratory, there is one big company which, as a 

general policy, requests the cleaning of weights 

together with a double calibration, before and after 

the cleaning. This company is in the pharmaceutical 

industry and is therefore strictly regulated. The 

management is used to issuing internal procedures 

and their staff is used to strictly following such 

procedures. 

We have monitored some of these calibrations 

over a period of approximately ten years.  

Although the SOP strictly advises the personnel 

not to touch the weights and use clean tools to 

handle the weights, these weights show clear signs 

of contamination. 

 
Figure 1: Typical calibration history (50 g, E2) blue 

(upper) line shows deviation from nominal before 

cleaning; red (lower) line are values after cleaning. 

Uncertainty bars 95 %. 

Figure 1 shows as an example a typical history of 

such a weight: the weight was first produced and 

calibrated in 2013 in a “clean” state. It was then re-

calibrated every two years with a calibration before 

(blue) and after (red) cleaning. In 2015 and 2017, 

the effect of cleaning is clearly visible: the values 

before cleaning are significantly higher. After 

cleaning, the values always come to the same stable 

level (within uncertainties). 

For the calibration of 2019, no contamination 

was detected: either the handling of the weight has 

changed significantly or the weight has not been 

used (this information was not available). 

The cleaning was done in a fully automated 

procedure with a heated bath with ultrasound 

cleaning using a regular, neutral industrial cleaner 

for stainless steel while constantly moving the 

weights. After cleaning, the weights are rinsed 

http://www.imeko.org/


ACTA IMEKO | www.imeko.org December 2020 | Volume 9 | Number 5 | 9 

multiple times in a cascade style process in filtered 

and de-ionised water and let stabilise and 

acclimatise according to [2]. The difference before 

and after cleaning of the same year is about -25 µg1 

for the 50 g weight, but varies between re-

calibrations. This means that the cleaning procedure 

has removed contamination of as much as 25 µg 

mass from the weight of 50 g nominal mass.  

There are two proofs that the cleaning did not 

remove any metal material from the weight (as one 

could suspect) but did effectively remove only dirt: 

first, all the “clean” values are consistent within 

their uncertainties. Second, we use the same type 

and material of weights in our calibration laboratory 

on robotic comparators as reference standard 

weights. For these weights, we do not observe any 

significant change in mass, neither over time nor 

before and after cleaning.  

We conclude, the weight pieces (would they be 

left un-touched) are stable with time and our 

cleaning procedure does not affect the pure metal 

mass of the weight pieces in any detectable way. 

7. IS THE CONTAMINATION 
SIGNIFICANT? 

Depending on the point of view, there are 

multiple answers to the question if such a 

contamination is significant. 

7.1. Significance considering Calibrated Values 
Referring to Figure 1, it is obvious that, since the 

uncertainty bars (95 %) of approximately 12 µg do 

not overlap, the difference in the values of 2015 and 

2017 are significant. En-values amount up to -1.6 

which, from the metrological view of the calibration 

laboratory clearly indicates that the values before 

and after cleaning are inconsistent and the amount 

of dirt removed therefore is significant. 

7.2. Significance if used “per Accuracy Class” 
One could also consider how the weights are 

used in the customer’s process: these weights are 

used for routine testing of balances. The 

requirements for the weights coming from the 

customer’s procedure is the adherence to class E2 

according to [2]. If the En-value is calculated using 

the MPE value of 100 µg instead of calibration 

uncertainty, En amounts to no larger than -0.2. Thus, 

it could be concluded that for the process of the 

application, the mass of contamination is 

insignificant since the weight value is allowed to 

vary within the class limits during its service life 

without affecting the balance test result. 

 
1 For the mass of removed contaminants as well as for 

correspondingly calculated En-values, we maintain 

negative values in this publication. 

7.3. Significance for Quality Control Interests 
The results are a clear sign that the instructions 

in the SOP (“handle weights with clean tweezers or 

with clean gloves”) are not effective or not 

effectively executed. Thus it can be concluded, that 

the mass of the contamination is significant. 

7.4. Significance if Trends are considered 
Should the balance testing results be analysed 

not only for passing or failing a limit test, but also 

for trends over time, then the contamination of the 

test weight might superpose the actual behaviour of 

the balance. Should trends in the order of size of the 

contamination be relevant, then the contamination 

might be significant. (The one example given in 

section 6 might be considered irrelevant if the 

weights are used e.g. on a balance with a readability 

of 10 µg and the amount of contamination is only a 

small multiple of the readability.) 

The considerations above underpin the general 

rule that only the user of the measurement device 

can make the final decision on the relevance of the 

data for his specific application. 

8. GENERAL VIEW 

8.1. En-Values of a Pool of Calibrations 
In our laboratory, we have observed some 

hundred calibrations of a pharmaceutical company 

with values before and after cleaning. The general 

picture is as follows: 

 
Figure 2: En-values vs. nominal mass 

As can be seen from Figure 2, we found many En-

values in the region between -1 and -2. As explained 

above, this is metrologically significant, but 

probably insignificant for the actual application. 

Nevertheless, there is a wide variation and there 

were some results where the cleaning resulted in En-

values greater than -3 and one even larger than -10 

(this was a 5 g weight). It is unknown unfortunately, 

how the customer reacted to these results.  

There are also some positive En-values (the 

weight values appeared heavier after cleaning). 

http://www.imeko.org/


ACTA IMEKO | www.imeko.org December 2020 | Volume 9 | Number 5 | 10 

These are owed to the reproducibility of the 

procedures of the calibration laboratory when 

obviously no significant contamination was present. 

8.2. Absolute Mass of Contaminants 
Another view on the pool of values is the 

absolute amount of contamination. Figure 3 shows 

the washed off contaminants’ mass vs. the nominal 

mass of the pieces. 

 
Figure 3: Absolute mass of contaminants vs. nominal 

mass 

It is obvious that with increasing mass (or 

probably rather with increasing surface), the mass 

of contaminants rises: a 1 mg piece shows hardly 

any contamination, while large contaminations 

occur for the largest weights. Furthermore, the more 

difficult to handle, the more contaminations occur. 

This holds especially true for weights of 2000 g and 

above.  

According to the impression of the author, this is 

owed to improper manual handling of the weights 

with improper (or even without) tools or 

inappropriate gloves. It is certainly not a 

metallurgical or corrosive surface effect since other 

weights of the same type under our own control 

behave absolutely stably (see section 6). 

9. POSSIBLE LOSS OF TRACEABILITY 

During a calibration, metrological traceability to 

the reference is established (definition 2.41 in [4]). 

During the service life of the calibration data (i.e. 

the calibration interval), traceability has to be 

maintained. This can be done by a combination of 

and not limited to: (1) intermediate checks using a 

check standard, (2) adding a component of 

uncertainty that accounts for assumed changes (e.g. 

𝑢𝑖𝑛𝑠𝑡  in [2]) or (3) a predictive modelling of the 
behaviour of the instrument. 

At the end of the calibration interval, a 

calibration must be performed in order to confirm 

that traceability has persisted throughout the 

interval. A possible means of doing this is an En-

test where the value and uncertainty assumed during 

service life are compared to value and uncertainty 

of the confirming end-of-interval calibration. If the 

confirmation of traceability fails, an according 

investigation is mandatory (e.g. “nonconforming 

work” in [3]). Possible consequences could be re-

considering the uncertainty assumptions or re-

considering the length of the calibration interval. 

It is obvious that for the example of weight 

pieces, the confirming calibration at the end of the 

interval must be done in a not-cleaned state, exactly 

in the state the weights were used until their last day 

in service. Cleaning of the weight piece and the 

following calibration are then used to again 

establish traceability for the interval to come. 

10. CONCLUSIONS 

From these findings, we draw the following 

conclusions: 

• The topic of cleaning of weights has been 
introduced historically, but is still legitimate 

today. 

• From an analysis of many weights from a highly 
regulated environment and used by trained 

personnel, we showed that contaminations occur 

frequently, even when there are clear 

instructions (SOPs) which should prevent 

contaminations. 

• Contaminations (in absolute units) tend to be 
large for weights of larger size. 

• Only the user of the weights – with deep 
metrological knowledge of the process in which 

the weights are being used – can answer the 

question if the amount of contamination is 

significant or relevant. 

• OIML R111-1 neglects the topic of calibrating 
before and after cleaning. 

• ISO 17025 is unclear (or incomplete) in listing 
cases in which a “before and after” calibration is 

mandatory and thus leaves the responsibility of 

the decision to the customer. 

• Metrological traceability is endangered by 
improper handling. Measures must be taken to 

maintain traceability during the calibration 

interval. 

• Calibrations before and after cleaning (“as 
found” and “as left”) are inevitable to gain all 

information necessary for appropriate 

management of weights as a metrologically 

traceable measuring device and to confirm 

traceability of the past interval. 

11. SUMMARY 

In this paper, we have presented the importance 

of cleaning weight pieces. Calibrations should be 

performed before and after cleaning in order to 

obtain information about the behaviour of the 

devices, about the effects the usage has, about the 

appropriateness of the uncertainty and of the 

calibration interval. The information from both 

calibrations is indispensable for establishing, 

http://www.imeko.org/


ACTA IMEKO | www.imeko.org December 2020 | Volume 9 | Number 5 | 11 

maintaining and confirming metrological 

traceability. 

12. REFERENCES 

[1] M. Stock et al., “Calibration campaign against the 
international prototype of the kilogram in 

anticipation of the redefinition of the kilogram 

part I: comparison of the international prototype 

with its official copies”, Metrologia, vol. 52, 

pp. 310-316, 2015. 

[2] OIML R111-1, Organisation Internationale de 
Métrologie Légale, Paris, France, 2004. 

[3] ISO/IEC 17025, International Organization for 
Standardization, Geneva, Switzerland, 2017. 

[4] JCGM 200, International vocabulary of metrology 
— Basic and general concepts and associated terms 

(VIM), BIPM et al., 2012.  

 
 

http://www.imeko.org/