Physica A 548 (2020) 123408

Contents lists available at ScienceDirect

Physica A
journal homepage: www.elsevier.com/locate/physa

Discussion

A power law in the ordering of the elements of the periodic

table
∗
Sergio Da Silva a , , Raul Matsushita b , Murilo Silva a
a
Department of Economics, Federal University of Santa Catarina, Brazil
b
Department of Statistics, University of Brasilia, Brazil

article info a b s t r a c t

Article history: We discover a power law in the periodic table between atomic number and atomic
Received 18 August 2019 weight that overlaps Mendeleev’s periodic law. Its Pareto exponent is computed as
Received in revised form 29 October 2019 1.0909. The power law can offer extra help in the quest for the next unknown element.
Available online 5 November 2019
© 2019 Elsevier B.V. All rights reserved.
Keywords:
Periodic table
Periodic law
Power law

1. Introduction

The periodic table lists the elements that compose all earthly substances, arranged in a way that reveals patterns.
Elements are fundamental substances from which all matter is made. Since the late 1600s, we know from Robert Boyle
that such elements are not air, earth, fire and water. In 1789, Antoine Lavoisier established a list of 33 elements and
classified them into groups — such as bromine, iodine, chlorine and lithium, sodium, potassium. Then, it was noted that
certain groups had similar properties, in terms of melting points and types of crystals formed. In the early 1800s, it was
realized that most atomic weights were multiples of the lightest element hydrogen. The atomic weight is the weight
of an atom of each element. From 1860 on, deeper patterns in the arrangements of atomic weights were discovered,
thanks to the insights from Stanislao Cannizzaro. Thus, the elements could be grouped by either their properties or atomic
weights. Dmitri Mendeleev’s breakthrough 150 years ago was to combine both in his periodic table. He arranged the
elements on one axis by atomic weight (in ‘‘periods’’) and on the other axis by properties, in groups. His periodic law
followed: as he ranked the elements by atomic weights, their chemical and physical properties recurred periodically. This
framework allowed him to notice that three elements were missing, and these were discovered soon after and named as
gallium, scandium and germanium [1]. Mendeleev’s generalization paved the way for the discovery of 55 more elements
since 1869. In 1913, Henry Moseley proved that we should consider atomic number rather atomic weight to rank the
elements [2]. Atomic number is the number of protons in each nucleus. However, the discovery of new elements still
fits Mendeleev’s classification system. Why? The physical explanation is that the mass of an atom is concentrated on its
nucleus, which is composed of protons and neutrons. However, because the number of neutrons in an element (isotopes)
can vary, it is still not completely clear why replacing Mendeleev’s atomic weight with Moseley’s atomic number does
not make much difference. This work offers such a missing justification.
Any prediction faces the age-old problem of induction. Because it is logically possible for the prediction to be false
while past evidence is true, the evidence does not conclusively establish the truth of the prediction [3]. The problem of
induction cannot be removed, so why has Mendeleev’s periodic law succeed? [4,5] Because it offers a tunnel to hopefully

∗ Correspondence to: Department of Economics, Federal University of Santa Catarina, Florianopolis SC 88049-970, Brazil
E-mail address: professorsergiodasilva@gmail.com (S. Da Silva).

https://doi.org/10.1016/j.physa.2019.123408
0378-4371/© 2019 Elsevier B.V. All rights reserved.
2 S. Da Silva, R. Matsushita and M. Silva / Physica A 548 (2020) 123408

Fig. 1. A power law in the periodic table.

track the next unknown elements. It converted unknown unknowns to known unknowns [6]. When we create a so-
called tunnel, we focus on a few well-defined sources of uncertainty [7], thus leaving out others that fall outside a law
domain. As stated by former U.S. Secretary Donald Rumsfeld, ‘‘There are known knowns. There are things we know that
we know. There are known unknowns. That is to say, there are things that we now know we don’t know. But there are also
unknown unknowns. There are things we do not know we don’t know’’. The conversion of unknown unknowns to known
unknowns has occurred repeatedly in chemistry through the formulation of hypotheses and experimental testing [8].
Here, we contribute to improving the roadmap given by the periodic table by showcasing a power law between atomic
number and atomic weight. The discovery of this power law explains that atomic number could replace atomic weight
with negligible prediction costs.
The periodic law takes into account chemical relations between the elements, of course. In contrast, a power law is a
regularity that is merely statistical. We conjecture that the power law describing the relationship between atomic number
n and atomic weight (abundance-weighted average of isotopes) w in the periodic table is w = a · nν , where w changes
as a power of n. We then quantify this power law [9] from the data of the current periodic table and calculate its Pareto
exponent ν [10,11].

2. Materials and methods

Few distributions follow a power law over their entire range — this emerges after a minimum threshold and sometimes
vanishes after a maximum threshold [10]. This is the reason that a distribution is referred to have a power law tail.
Considering that our conjectured power law becomes log w = log a + ν log n after taking base 10 logarithms (any base
will work), a straight-line power law form reveals itself only after a minimum threshold nmin . Pareto exponent ν essentially
means that, relative to a given atomic number n, finding more elements in the periodic table is proportional to nν [12,13].
Pareto exponent tracks the degree of uncertainty of repeating the discovery of a new element after a certain threshold
atomic number n.

3. Results

Table 1 displays the current periodic table of 118 elements, while Fig. 1 showcases an example for n ≥ 6. A power
law reveals itself in a straight line of a log–log plot of the previous equation, where log a is the y-intercept = d, and ν is
the slope. Thus, a = 10d . From the fitting line in Fig. 1, we find
log w = 0.1994 + 1.1028 · log n,
where the estimates with their corresponding standard errors are ν = 1.1028± 0.0027 (slope), d = 0.1994± 0.0047
(intercept), and then a = 100.1994 = 1.5826. The implied power law is then
w = 1.5826n1.1028 .
To illustrate, the atomic weight of the next unknown element n = 119 is casually predicted as
w = 1.5826 × 1191.1028 ≈ 307.8.
This graphical exercise can be complemented by an alternative nonlinear least squares’ approach [9]. Here, we find
the power law:
w = 1.6683n1.0909 ,
 S. Da Silva, R. Matsushita and M. Silva / Physica A 548 (2020) 123408 3

Table 1
Current periodic table.
Source: Royal Society of Chemistry
Element Rank, n Atomic weight, w log n log w
Hydrogen 1 1.008 0 0.00346
Helium 2 4.0026 0.30103 0.60234
Lithium 3 6.94 0.47712 0.84136
Beryllium 4 9.0122 0.60206 0.95483
Boron 5 10.81 0.69897 1.03383
Carbon 6 12.011 0.77815 1.07958
Nitrogen 7 14.007 0.8451 1.14635
Oxygen 8 15.999 0.90309 1.20409
Fluorine 9 18.998 0.95424 1.27871
Neon 10 20.18 1 1.30492
Sodium 11 22.99 1.04139 1.36154
Magnesium 12 24.305 1.07918 1.3857
Aluminum 13 26.982 1.11394 1.43107
Silicon 14 28.085 1.14613 1.44847
Phosphorus 15 30.974 1.17609 1.491
Sulfur 16 32.06 1.20412 1.50596
Chlorine 17 35.45 1.23045 1.54962
Argon 18 39.948 1.25527 1.6015
Potassium 19 39.0983 1.27875 1.59216
Calcium 20 40.078 1.30103 1.60291
Scandium 21 44.956 1.32222 1.65279
Titanium 22 47.867 1.34242 1.68004
Vanadium 23 50.942 1.36173 1.70708
Chromium 24 51.996 1.38021 1.71597
Manganese 25 54.938 1.39794 1.73987
Iron 26 55.845 1.41497 1.74698
Cobalt 27 58.933 1.43136 1.77036
Nickel 28 58.693 1.44716 1.76859
Copper 29 63.546 1.4624 1.80309
Zinc 30 65.38 1.47712 1.81544
Gallium 31 69.723 1.49136 1.84338
Germanium 32 72.63 1.50515 1.86112
Arsenic 33 74.922 1.51851 1.87461
Selenium 34 78.971 1.53148 1.89747
Bromine 35 79.904 1.54407 1.90257
Krypton 36 83.798 1.5563 1.92323
Rubidium 37 85.468 1.5682 1.9318
Strontium 38 87.62 1.57978 1.9426
Yttrium 39 88.906 1.59106 1.94893
Zirconium 40 91.224 1.60206 1.96011
Niobium 41 92.906 1.61278 1.96804
Molybdenum 42 95.95 1.62325 1.98204
Technetium 43 98 1.63347 1.99123
Ruthenium 44 101.07 1.64345 2.00462
Rhodium 45 102.906 1.65321 2.01244
Palladium 46 106.42 1.66276 2.02702
Silver 47 107.87 1.6721 2.0329
Cadmium 48 112.414 1.68124 2.05082
Indium 49 114.818 1.6902 2.06001
Tin 50 118.71 1.69897 2.07449
Antimony 51 121.76 1.70757 2.0855
Tellurium 52 127.6 1.716 2.10585
Iodine 53 126.904 1.72428 2.10348
Xenon 54 131.293 1.73239 2.11824
Cesium 55 132.905 1.74036 2.12354
Barium 56 137.327 1.74819 2.13776
Lanthanum 57 138.905 1.75587 2.14272
Cerium 58 140.116 1.76343 2.14649
(continued on next page)

where the standard errors for a and ν are, respectively, 0.036 and 0.0048. The Pareto exponent is thus statistically different
from 1. Now, the atomic weight of the next unknown element n = 119 is predicted as

w = 1.6683(119)1.0909 ≈ 306.54.
Considering a Monte Carlo-based 95-percent prediction interval for this nonlinear regression, after 106 replications, we
find [300.78; 312.20]. This is the tunnel for predicting what is offered by the power law.
4 S. Da Silva, R. Matsushita and M. Silva / Physica A 548 (2020) 123408

Table 1 (continued).
Element Rank, n Atomic weight, w log n log w
Praseodymium 59 140.908 1.77085 2.14894
Neodymium 60 144.242 1.77815 2.15909
Promethium 61 145 1.78533 2.16137
Samarium 62 150.36 1.79239 2.17713
Europium 63 151.964 1.79934 2.18174
Gadolinium 64 157.25 1.80618 2.19659
Terbium 65 158.925 1.81291 2.20119
Dysprosium 66 162.5 1.81954 2.21085
Holmium 67 164.93 1.82607 2.2173
Erbium 68 167.259 1.83251 2.22339
Thulium 69 168.934 1.83885 2.22772
Ytterbium 70 173.045 1.8451 2.23816
Lutetium 71 174.967 1.85126 2.24296
Hafnium 72 178.49 1.85733 2.25161
Tantalum 73 180.948 1.86332 2.25755
Tungsten 74 183.84 1.86923 2.26444
Rhenium 75 186.207 1.87506 2.27
Osmium 76 190.23 1.88081 2.27928
Iridium 77 192.217 1.88649 2.28379
Platinum 78 195.084 1.89209 2.29022
Gold 79 196.967 1.89763 2.29439
Mercury 80 200.592 1.90309 2.30231
Thallium 81 204.38 1.90849 2.31044
Lead 82 207.2 1.91381 2.31639
Bismuth 83 208.98 1.91908 2.3201
Polonium 84 209 1.92428 2.32015
Astatine 85 210 1.92942 2.32222
Radon 86 222 1.9345 2.34635
Francium 87 223 1.93952 2.3483
Radium 88 226 1.94448 2.35411
Actinium 89 227 1.94939 2.35603
Thorium 90 232.038 1.95424 2.36556
Protactinium 91 231.036 1.95904 2.36368
Uranium 92 238.029 1.96379 2.37663
Neptunium 93 237 1.96848 2.37475
Plutonium 94 244 1.97313 2.38739
Americium 95 243 1.97772 2.38561
Curium 96 247 1.98227 2.3927
Berkelium 97 247 1.98677 2.3927
Californium 98 251 1.99123 2.39967
Einsteinium 99 252 1.99564 2.4014
Fermium 100 257 2 2.40993
Mendelevium 101 258 2.00432 2.41162
Nobelium 102 259 2.0086 2.4133
Lawrencium 103 262 2.01284 2.4183
Rutherfordium 104 267 2.01703 2.42651
Dubnium 105 268 2.02119 2.42813
Seaborgium 106 269 2.02531 2.42975
Bohrium 107 270 2.02938 2.43136
Hassium 108 269 2.03342 2.42975
Meitnerium 109 278 2.03743 2.44404
Darmstadtium 110 281 2.04139 2.44871
Roentgenium 111 280 2.04532 2.44716
Copernicium 112 285 2.04922 2.45484
Nihonium 113 286 2.05308 2.45637
Flerovium 114 289 2.0569 2.4609
Moscovium 115 289 2.0607 2.4609
Livermorium 116 293 2.06446 2.46687
Tennessine 117 294 2.06819 2.46835
Oganesson 118 294 2.07188 2.46835

A final caution note is worthwhile. Consider the practical implication of rounding ν = 1.0909 to either 1 (a so-called
Zipf’s law) or 1.1. We warn this rounding is highly sensitive and the prediction error escalates as a result [14].

4. Conclusion

Mendeleev’s periodic law offers a tunnel to predict unknown elements from the periodic table. Replacing Mendeleev’s
atomic weight with Moseley’s atomic number does not make much difference for such a prediction. We discover a power
 S. Da Silva, R. Matsushita and M. Silva / Physica A 548 (2020) 123408 5

law between atomic number and atomic weight that justifies this, and compute its Pareto exponent as ν = 1.0909. The
power law overlaps the periodic law and contributes to refine tunneling in the search for a new element by converting
unknown unknowns to known unknowns.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have
appeared to influence the work reported in this paper.

Acknowledgments

The authors thank the useful comments made by anonymous reviewers.

Funding

The authors acknowledge the financial support from CNPq, Brazil, CAPES, Brazil and FAPDF, Brazil.

References

[1] E.R. Scerri, The evolution of the periodic system, Sci. Am. 279 (3) (1998) 78–83.
[2] E. Scerri, The past and future of the periodic table, Am. Sci. 96 (2008) 52–58.
[3] M.R. Foster, Prediction and the problem of induction. Semantic Scholar, 1998, pp. 1–8, https://pdfs.semanticscholar.org/9a97/
4d04be191df6779b60e54e2bf7cf3287a49a.pdf.
[4] P. Armbruster, F.P. Hessberger, Making new elements, Sci. Am. 279 (3) (1998) 72–77.
[5] E. Scerri, Cracks in the periodic table, Sci. Am. 308 (6) (2013) 68–73.
[6] R. Matsushita, S. Da Silva, R. Da Fonseca, et al., Bypassing the truncation problem of truncated Lévy flights, submitted for publication.
[7] N.N. Taleb, The Black Swan: The Impact of the Highly Improbable, Random House, New York, 2010.
[8] H.D. Loxdale, B.J. Davis, R.A. Davis, Known knowns and unknowns in biology, Biol. J. Linn. Soc. 117 (2) (2016) 386–398.
[9] X. Gabaix, Power laws in economics: An introduction, J. Econ. Perspect. 30 (1) (2016) 185–206.
[10] M.E. Newman, Power laws, Pareto distributions and Zipf’s law, Contemp. Phys. 46 (5) (2005) 323–351.
[11] A. Clauset, C.R. Shalizi, M.E. Newman, Power-law distributions in empirical data, SIAM Rev. 51 (4) (2009) 661–703.
[12] S. Da Silva, M. Perlin, R. Matsushita, et al., Lotka’s law for the Brazilian scientific output published in journals, J. Inf. Sci. 45 (5) (2019) 705–709.
[13] M. Kawamura, C.D. Thomas, A. Tsurumoto, et al., Lotka’s law and productivity index of authors in a scientific journal, J. Oral Sci. 42 (2) (2000)
75–78.
[14] S. Da Silva, R. Matsushita, R. Giglio, et al., Granularity of the top 1000 Brazilian companies, Physica A 512 (C) (2018) 68–73.