The New SI and the CODATA recommended values of the fundamental

constants 2014 (arXiv:at-phys-1507.07956)

Franco Pavese
formerly National Research Council, Istituto di Metrologia G.Colonnetti (from 2006, Istituto
Nazionale di Ricerca Metrologica), Torino, Italy
e-mail: frpavese@gmail.com
This notes aim is to point out some standing features of the present CODATA method in the
light of the recent CODATA table of the 2014 recommended values for the fundamental
constants published on paper arXiv:1507.07956. A comprehensive discussion on this and
related issues is becoming very important in view of the foreseen revision of the SI, presently
planned for 2018. These features may raise doubts on a possible mixing of physical reasons of
general validity in science with some needs specific of metrology concerning the base units of
the International System (SI) of measurement units.

Introduction
In an article on arXiv:1507.07956 dated July 30, 2015, the CODATA Group on fundamental
constants published a summary of the results of their 2014 adjustment of the fundamental constants,
[1] following the full publication in 2012 of the 2010 adjustment [2]. In the mean time, the
Confrence Gnrale des Poids et Measures examined in 2011 and 2014 the proposal presented by
the BIPM to deeply revise the definitions of the present base units of the SI by using the numerical
values of constants in their definition. [3]
As a long since metrologist, for some years I am observing this process, and in particular the
CODATA numerical outcomes, obtained with the application of the Least Squares Adjustment
(LSA) method, which are normally used in the official BIPM documents as the ones to be used to
define the numerical values of those constants in the unit definitions. In 2014 a Letter devoted to
this CODATA issue was published [4], and a Reply to it too [5] on the same Journal.
The first thing that I noticed in [1], is that in 2014 there are changes in the numerical values of the
five constants that are relevant to the present proposal concerning the SI. The second is that
CODATA, in applying its LSA method, still kept fixed not only 0, but also c0, and thus also 0.
I will not take into account in this Note the effect of possible changes of the BIPM proposal
following the recently published paper arXiv:1510.08324, [6] nor the possible implications of
another recent paper, arXiv:1410.7906v1. [7]
My comments will concern both the above issues, and some possible consequences of them.
Illustration of the observed problems
Changes in the recommended values of the constants
In paper [1] one can observe that the adjusted values of the relevant constants have still changed
since 2010. The paper only discusses the CODATA uncertainties. However, for the New SI, where
the numerical values are supposed to be then stipulated, also the values being the best in a strict
statistical sense are an important issue. Table 1 and Fig. 1 report a comparison of the 20062010
data (c0 being fixed). Also mu is included, considered in [6].

Table 1. Change in numerical value of the CODATA adjustments 2006-2014. [1, 6]

Constant
e 1019
h 1034
NA 1023
kB 1023
mu 1027

CODATA

Num. value *

s.d. 107

Change

2006
2010
2014
2006
2010
2014
2006
2010
2014
2006
2010
2014
2006
2010
2014

1.602 176 49
1.602 176 57
1.602 176 62
6.626 0690
6.626.0696
6.62607004
6.022 1418
6.022 1413
6.022 1409
1.380 6504
1.380 6488
1.380 6485
1.660 538 78
1.660 538 92
1.660 539 04

0.4
0.35
0.1
3.3
2.9
0.8
3
2.7
0.7
24
13
8
0.8
0.7
0.2

8 108
5 108

6 107
8 107

5 107
4 107

1.6 106
0.3 106

12 108
14 108

Total shift
1.3 107
1.4 106
9.0 107
1.9 106
2.6 107

* The smaller-case digit is uncertain.

Fig. 1. Behaviour with time of the CODATA changes in value, with uncertainty bars. [1, 6]

The trend of change of e, h and NA (and mu) does not indicate that one can be confident that the
2014 one might be the last significant variation.
On the other hand, the 2014 values for e, h and NA (and mu) are clearly not metrologically
compatible (for k = 1) with the 2006 value.

It will be necessary to wait for the publication of the full paper on the 2014 adjustments, especially
for understanding the result of the critical review of the new results and have the list of the
experimental data admitted to the adjustment database, to draw final conclusions.
Meaning and way to use the LSA method
The LSA method needs one reference value in the analysed set, in the absence in this case of an
additional condition that could remove the over-determinacy of the set of equations. This is due to
the fact that one does not know the true numerical value of any constant. Thus, the use of a fixed
(not adjusted) value is necessary, and sufficient, meaning that all adjustments are relative to this
value: the numerical value of 0 was originally used. Should its value be found to have changed, or
should another member of the set be chosen to stand fixed, all the values of the numerical
adjustments would change accordinglyonly the pair differences remaining unchanged.
On the other hand, since 1983, when one constant, c0, was stipulated in the SI definition of the unit
of length, CODATA used for their analyses that fixed value also for c0, and consequently for 0.
Since then, the numerical values of these three constants are indicated as exact in the tables. [1]
When having a single fixed member of the set, the bias induced by its choice is irrelevant in many
other fields where the goal is only to check for the best consistency of the set. In the case of the use
of the constants in measurement units, instead, also the continuity in magnitude of the base units
through the change of the unit definitions is a must, so that the adjusted values should considered to
also optimising that continuity. This may place constraints on the choice of the fixed member,
possibly not even being 0 the best one fitting the purpose. A study on this issue, specific of
metrology, is not yet available.
It is a well-known fact that the LSA causes an intimate relationship which exist [is set by the LSA]
among least-squares adjusted values of the fundamental constants, so that a significant shift in
the numerical value of one will generally cause significant shifts in others [8]. When having more
than one fixed value, it becomes problematic to understand the effect and the level of the bias
induced by this fact. A published study on the changes in the adjustments posterior to 1983, in
respect to leaving c0, and thus also 0, adjustable or not is not available.
The three previous issues are metrologically relevant. However, on the other hand, the CODATA
analyses are of the greatest importance in a much wider context. Therefore, a too large influence of
metrological constraints into the normal decisions of CODATA in the application of the LSA
method (wider in usefulness than it is for the metrological issues) and in the resulting adjustments,
should not be seen favourably by scientists in general.
In this respect, what happened in 1973 is exemplary of something that should not have happened.
As reported in [4] The 1973 adjustment preserved the value c0 = 299 792 458 m s1 (Without
intending to prejudge any future redefinition of the metre or the second, the CCDM suggested that
any such redefinitions should attempt to retain this value provided that the data upon which it is
based are not subsequently proved to be in error [9]), but not the uncertainty, set to 1.2 m s1
(4109 relative). The value, and its uncertainty, used instead for the 1973 LSA, as said in [9], was
the Evensons one [10], (299 792 456.2 1.1) m s1 (3.5109 relative). Notice than Mulligan [11]
reports for the same Evenson work the value (299 792 457.4 1.1) m s1. Later, Blaney et al. [12]
in 1974 obtained 299 792 459.0(0.8) m s1. Apparently, the latter determination was not
considered in the 1983 stipulation of c0.
The issue already cited, that the CODATA kept fixed c0, and thus also 0, is an improper bias on the
whole set. With new measurements of other constants, it could have happened that the LSA method
supplied a different value for c0, an outcome perfectly normal for the LSA procedure and
acceptable in general science. However, the stipulation is only relevant to the unit of length, where

the stipulated value can stand fixed to the original fixed value (with a proper wording of the
definition) even for ever. Keeping instead c0 and 0 fixed in the study of other features of the
physical laws may alter the adjustments of other constants. Also this study is not yet available.
That type of constraint would be dramatically more influent on science after stipulation of the
multiple constants needed to the definition of the New SI, which not only would fix the value of the
constants appearing in the definitions, but will also have consequences on others functionally
depending on them.
These issues need to be clarified in a convincing way. Science in general, should not be tied to these
constraints.
Conclusions
In the last decade an extraordinary scientific effort, theoretical and experimental, has been
performed for measuring a number of fundamental constants on stronger bases and with decreased
uncertainty, bringing to extraordinary results. However, one should never forget a risk that past
experience has shown to be less than unexpected, and that is considered in the usual state of affairs
by scientists working on human factors [13]: the fatal attraction of a target value or threshold
uncertainty value.
In this situation, special provisions should be set to increase confidence that this occurrence is
prevented, and that the present outcomes, namely the persistence in time of the stipulated values,
can be checked in future. Authoritative scientists may agree, I think, on the fact that science has no
fixed deadlines, outside facts that bring to a convincing inter-subjective decision of considering the
goal achieved at its besti.e., to be sound and to fit the purpose for the scientists and for the users,
in a given moment of science history. Actually, decisions taken once and for ever are not typical
of science.
1. P.J. Mohr, D.B. Newell and B.N. Taylor, CODATA recommended values of the fundamental constants
2014, (2015) arXiv:1507.07956v1.
2. P.J. Mohr, B.N. Taylor and D.B. Newell, CODATA Recommended Values of the Fundamental Physical
Constants: 2010, Rev. Modern Phys. 84 (2012) 194.
3. http://www.bipm.org/en/measurement-units/new-si/
4. F. Pavese, Some problems concerning the use of the CODATA adjusted values of fundamental constants
in the definition of measurement units, Metrologia 51 (2014) L1-L4.
5 E. de Mirandes, Reply to Some problems concerning the use of the CODATA adjusted values of
fundamental constants in the definition of measurement units, Metrologia 51 (2014) L5-L7.
6. N. Fletcher, R.S. Davis, M. Stock and M.J.T. Milton, Modernizing the SI implications of recent progress
with the fundamental constants, arXiv 1510.08324 (October 15, 2015).
7. K.A. Bronnikov, V.D. Ivashchuk, M.I. Kalinin, V.V. Khruschov, S.A. Kononogov and V.N. Melnikov,
On new definitions of SI base units. Why is the atomic kilogram preferable, Measurement Techniques
(english), May 2015, DOI 10.1007/s11018-015-0674-6; arXiv:1410.7906v1 29 October 2014.
8. B.N. Taylor, W.H. Parker and D.N. Langenberg, Rev. Mod. Phys. 41 (1969) 375496; D.N. Langenberg
and B.N. Taylor (ed), Precision Measurement and Fundamental Constants (NBS Special Publication 343)
(Washington, DC: NBS) 1970
9. E.R. Cohen and B.N. Taylor, The 1973 Least-Squares Adjustment of the Fundamental Constants, J. Phys.
Chem. Ref. Data 2 (1973) 663734
10. K.M. Evenson, J.S. Wells, F.R. Petersen, B.L. Danielson and G.W. Day, R.L. Barger and J.L. Hall,
Speed of light from direct frequency and wavelength measurements of the methane-stabilized laser, Phys.
Rev. Lett. 29 (1972) 13461349
11. J.F. Mulligan, Some recent determinations of the velocity of light. III, Am. J. of Physics 44 (1976) 960
969

12, T.G. Blaney, C.C. Bradley, G.J. Edwards, D.J.E. Knight, W.R.C. Eowley, K.C. Shotton and P.T. Woods,
Ann. N.Y. Acad. Sci 251 (1974) 46
13. M. Henrion and B. Fischhoff, Assessing Uncertainty in Physical Constants, Ch. 9 in Judgement and
Decision Making (B. Fischhoff, Ed.) Routhledge (2013) pp. 172-187, 113649734X, 9781136497346, taken
from American Journal of Physics 54, 791 (1986); doi: 10.1119/1.14447