Paulis Exclusion Principle PDF
Paulis Exclusion Principle PDF
Paulis Exclusion Principle PDF
The exclusion principle was the final outcome of Pauli’s struggle to under-
stand some spectroscopic anomalies in the early 1920s: doublets were
observed in the spectra of alkali metals, singlets and triplets in the spectra
of the alkaline earths, and even more anomalous patterns were observed
when chemical elements were placed in an external magnetic field (anom-
alous Zeeman effect and Paschen–Back effect). These anomalous spectra
challenged the old quantum theory, and prompted a radical theoretical
change (Section 2.1). From 1920 to 1924 Alfred Landé, Werner
Heisenberg, and Niels Bohr were all engaged in trying to save the trad-
itional spectroscopic model (the so-called atomic core model) and to
reconcile it with the observed anomalies. The impasse was solved only
with Pauli’s introduction of a fourth degree of freedom for the electron,
and the consequent demise of the atomic core model (Section 2.2). What
Pauli called the ‘twofoldness’ [Zweideutigkeit] of the electron’s angular
momentum was soon reinterpreted as the electron’s spin (Section 2.3).
Pauli’s exclusion rule was announced in this semi-classical spectroscopic
context that characterized the revolutionary transition from the old quan-
tum theory to the new quantum theory around 1925.
35
36 2 The origins of the exclusion principle
almost eighty years – the time necessary to improve the quality of technical
instruments – to find an empirical law governing the line distribution in the
hydrogen spectrum as observed in stars. In 1885 Balmer discovered that
each of the fourteen lines composing the spectrum of hydrogen obeyed the
simple rule
n22
l¼h (2:1)
n22 n21
R R
n ¼ 2
(2:2)
ðn1 þ 1 Þ ðn2 þ 2 Þ2
where n is the frequency of the n-th member of the series, R is the Rydberg
constant, and denotes spectral terms (e.g. 1 ¼ 3P1, 2 ¼ 3S1). By setting
1 ¼ 0, 2 ¼ 0, n1 ¼ 2 and n2 ¼ 3, 4, 5, . . . Rydberg’s formula reduces to
Balmer’s.
Rydberg was the first to distinguish between a sharp series (S) and a
diffuse series (D). Other types of series were later discovered: the so-called
principal series (P) and the fundamental series (F). Jointly they form the
four chief series (S, P, D, F) available for every type of line (i.e. singlet,
doublet, triplet, . . . ).1 According to Rydberg’s formula, each line of a
series is then given by the difference between two spectral terms.2 This
was a useful phenomenological law, but did not offer a theoretical under-
standing of spectra. Only with Bohr’s atomic theory of hydrogen in 1913,
was the first step towards such an understanding taken.
Not only could Bohr derive from his atomic theory Rydberg’s and
Balmer’s formulae for the hydrogen series, but he provided Rydberg’s
1
The standard notation uses the superscript on the left of the capital letter to denote the type
of line; for instance the singlet four chief series are denoted as 1S, 1P, 1D, 1F.
2
For instance, the three lines composing a triplet of the sharp series S are given by the
following differences: 3P2 3S1; 3P1 3S1; 3P0 3S1, where the subscripts indicate the
specific line of the series at issue (e.g. 3P2 denotes the second line of the principal series
triplet). These differences are obtained by plugging into Rydberg’s formula 2 ¼ 3S1 (fixed
term) and 1 ¼ 3P2, 3P1, 3P0 respectively (running terms).
2.1 The prehistory of Pauli’s exclusion principle 37
where k and r are respectively the azimuthal and the radial quantum
number and, following Bohr’s quantum condition, they take only discrete
integral values (r ¼ 0, 1, 2, 3, . . . ; k ¼ 1, 2, 3, 4, . . . ). The sum r + k was
equal to n, where n was the so-called principal quantum number denoting
the energy-state of the electron. Like r and k, n also took only integral
values (n ¼ 1, 2, 3, . . . ). While r and k determined respectively the size and
shape of the orbit, in order to determine the orientation of the orbit in an
external magnetic field, it was necessary to introduce a third degree of
freedom.
3 4
Sommerfeld (1916a), (1919). For details see White (1934), pp. 42ff.
38 2 The origins of the exclusion principle
5
Larmor (1897). 6 Sommerfeld (1916b).
7
It was assumed that m could not take the value 0, because this would lead to the electron
colliding with the nucleus.
2.1 The prehistory of Pauli’s exclusion principle 39
k=3
m=+3
k=2
m = +2 +2
k=1
m = +1 +1 +1
2
k=1
k=
–1 –1 –1
–2 –2
–3
8
The Stern–Gerlach experiment consisted of directing a beam of silver atoms in a high
vacuum through collimating slits along a strong field gradient so that the beam split into
two sub-beams. According to classical mechanics, since the magnetic moments of the atoms
were isotropically distributed (i.e. they had equal probability to take on any value
between M and –M), one would expect one single spot at the centre of the distribution
corresponding to the average magnetic moment. But Stern and Gerlach observed two spots
centred at two different points corresponding to two discrete values of M. This result was
greeted as a confirmation of Sommerfeld’s space quantization, as evidence for the discrete
orientations atoms would take up in the presence of an external magnetic field. As a matter
of fact, the Stern–Gerlach experiment can be unambiguously interpreted only with the
introduction of the spin: the correct explanation is that the total angular momentum of the
silver atoms coincides with the spin of the valence electron (which is alone responsible for
the magnetic moment M of the atom). And the spin projection takes only two discrete
eigenvalues (+1/2 and 1/2 in units of h / 2p) corresponding to the two observed spots.
40 2 The origins of the exclusion principle
9
See Bohr (1923a). Reprinted in Bohr (1977), p. 632.
10
‘Although the process of radiation cannot be described on the basis of the ordinary theory
of electrodynamics, according to which the nature of the radiation emitted by an atom is
directly related to the harmonic components occurring in the motion of the system, there is
found nevertheless to exist a far-reaching correspondence between the various types of
possible transitions between the stationary states on the one hand and the various
harmonic components of the motion on the other. This correspondence is of such a
nature that the present theory of spectra is in a certain sense to be regarded as a rational
generalization of the ordinary theory of radiation.’ Bohr (1914). Reprinted in
Bohr (1976), p. 301.
11
For a detailed analysis of the implications of the correspondence principle for Bohr’s later
research as well as for the development of quantum mechanics, see Petruccioli (1988).
2.1 The prehistory of Pauli’s exclusion principle 41
12
The concept of adiabatic change goes back to Boltzmann’s and Clausius’s attempts to
reduce the second law of thermodynamics to pure mechanics, while the notion of adiabatic
motion was developed by Hertz and Helmoltz.
13
Bohr (1923a). Reprinted in Bohr (1977), pp. 631–2.
14
See for instance Serwer (1977), p. 204.
15
Ehrenfest (1913) was the first to extend the classical notion of adiabatic invariants to
quantum theory as well as the first to introduce – following a terminology suggested by
Einstein – an adiabatic principle. For the history of this principle, see Jammer (1966),
second edition (1989), pp. 99–107. It is worth noticing that Bohr derived the invariance of
statistical weights precisely from Ehrenfest’s adiabatic principle: ‘A basis for the
42 2 The origins of the exclusion principle
treatment of this problem [i.e. the problem of the weights that must be assigned to the
individual stationary states] is contained in a general condition, derived by Ehrenfest . . .
stating that in a continuous transformation of a system, the statistical weights of the
individual stationary states do not change as long as the degree of periodicity remains the
same. For systems for which the degree of periodicity is equal to the number s of the
degrees of freedom, this condition involves that all stationary states must be assigned the
same weight hs.’ Bohr (1923a). Reprinted in Bohr (1977), p. 616.
16
This is particularly evident in the electronic distribution of neon (Z ¼ 10), where the ten
electrons were distributed as follows: the first two electrons in two (11) orbits, the following
four electrons in four (21) orbits and the last four electrons in four (22) orbits. For a detailed
discussion of Bohr’s schema and its drawback, see Heilbron (1982), p. 273.
17
For the discovery of hafnium, see Kragh (1979).
2.1 The prehistory of Pauli’s exclusion principle 43
18
For instance k ¼ 1 corresponds to s; k ¼ 2 to p; k ¼ 3 to d and k ¼ 4 to f. Notice the
notational difference between the small letters s, p, d, f referring to the energy-levels (and
hence to the spectral terms) for individual electrons and the capital letters S, P, D, F
referring to the spectral terms resultant from the combination of any two or more
non-equivalent s, p, d, f electrons.
44 2 The origins of the exclusion principle
19
For a historical reconstruction of the research on the X-rays and their spectra see Heilbron
(1966), (1967).
20
Sommerfeld (1916a). 21 See de Broglie and Dauvillier (1922).
22
See Forman (1968). 23 Sommerfeld (1920). 24 Landé (1921a), (1921b).
2.1 The prehistory of Pauli’s exclusion principle 45
25
Rubinowicz (1918). Rubinowicz was one of Sommerfeld’s students and assistants. While
working on the conservation of angular momentum in the classical process of radiation, he
found the following integral selection rules for the azimuthal and the magnetic quantum
numbers: k ¼ 0, 1 and m ¼ 0, 1. But Rubinowicz identified the azimuthal quantum
number with the total angular momentum of the atom, rather than with the electron’s
orbital angular momentum. Accordingly, the physical meaning of this rule was that the
total angular momentum of the atom was quantized as required by Sommerfeld’s space
quantization. For the historical reconstruction of the way Landé was driven to identify
j with the total angular momentum via Rubinowicz’s selection rule, see Forman (1970),
p. 196.
46 2 The origins of the exclusion principle
There were however a few striking difficulties with the atomic core
model. First, it contradicted Bohr’s building-up principle. Consider, for
instance, the atomic cores of alkalis. They coincide with the noble gases
occupying the column immediately preceding them in Mendeleev’s peri-
odic table. Hence, if alkalis had a non-zero core momentum as the model
required, noble gases too should have a non-zero angular momentum. But
this was in contrast with experimental evidence.
Furthermore, even granted a non-zero angular momentum for noble
gases, it was still impossible to get doublets and, more generally, even
multiplets. According to the atomic core model, the values of the inner
quantum number j were defined as | k r | j | k + r |. But given the
integral values of the electron azimuthal quantum number k and the core
quantum number r, it was possible to get spectral terms of odd multiplicity
(e.g. singlets, triplets, . . . ), but not of even multiplicity (e.g. doublets,
quadruplets, . . . ). Alkali doublets remained unexplained.
The atomic core model contradicted Bohr’s building-up principle in
another way: the inner quantum number j and the number of states
associated with it did not remain adiabatically invariant during the build-
ing-up process. As Landé26 pointed out, the inner quantum number j+ of
an ion did not coincide with the core quantum number r of the immediately
preceding element in the periodic table, as Bohr’s principle required.
Rather, Landé found that
j þ ¼ r 1=2 (2:9)
and more generally
j ¼ r þ k 1=2; r þ k 3=2; : : : (2:10)
The number of states associated with j seemed to decrease in the building-
up process by an inexplicable half-integral unit. And this decrease brought
with it a corresponding decrease in the statistical weights of the states. But
the building-up principle required the statistical weights to remain invari-
ant. The riddle of statistical weights 27 was no less puzzling than the
doublet riddle. Indeed they were two sides of the same coin as the analysis
of the Zeeman effect soon revealed.
26
Landé (1923a) See also Serwer (1977), pp. 206–7.
27
For this terminology see Serwer (1977), pp. 204–7.
2.1 The prehistory of Pauli’s exclusion principle 47
Y Y Y
(a) (b) (c) (d)
Y
+
x
+
y– X+ y–
X y+ X X
x
X–
y+
–
x–
Z Y Z Z H
⊥ ⊥
ν0
ν
–∆ν +∆ν
−∆
+∆
–∆ν +∆ν
ν
ν0
X
s s s ν0 s
H
P
ν0 ν
Z (e) (A) (B)
Fig. 2.2 Diagrams for Lorentz’s explanation of the normal Zeeman effect,
where (A) is the diagram for the triplet and (B) for the doublet.
Source: H. E. White (1934) Introduction to Atomic Spectra (New York,
London: McGraw-Hill Book Company). Reproduced with permission of
The McGraw-Hill Companies.
28 29
Zeeman (1896). Lorentz (1897).
48 2 The origins of the exclusion principle
No field
Weak field
Fig. 2.3 Spectral patterns of the normal Zeeman effect of zinc and of the
anomalous Zeeman effect of sodium, viewed perpendicular to the magnetic
field.
Source: H. E. White (1934) Introduction to Atomic Spectra (New York,
London: McGraw-Hill Book Company). Reproduced with permission of
The McGraw-Hill Companies.
30
Runge (1907).
31
For instance, the anomalous Zeeman pattern of the sodium principal series doublet can be
expressed as 2/3L for the quadruplet and 4/3L for the sextuplet, where L is the Lorentz
unit of the Zeeman effect L ¼ (eH/(4pmc2)).
2.1 The prehistory of Pauli’s exclusion principle 49
derive the anomalous doublets, but at the cost of strengthening the doublet
riddle and the riddle of statistical weights.
Landé32 empirically ascribed half-integral values to the magnetic quan-
tum number (and hence also to the inner quantum number j): m ¼ 1/2,
3/2, 5/2, . . . , ( j 1/2) so that he could derive doublets. But half-
integral values for j and m clearly violated the Bohr–Sommerfeld theory,
which allowed only integral values for quantum numbers, as well as Bohr’s
building-up principle, which forbade the decrease of a half-integral unit in
the building-up process. The riddle of statistical weights was still looming
on the horizon. Furthermore, half-integral quantum numbers for alkali
doublets were in contrast with relativistic X-ray doublets that required
only integral quantum numbers: Landé’s proposal strengthened the doub-
let riddle.
Landé also calculated the energy W of a Zeeman state (i.e. the Zeeman
Hamiltonian describing the interaction energy between the external mag-
netic field and the magnetic moment of the atom) as
W ¼ W0 þ ½mghðeH=4pmcÞ ¼ W0 þ mghL (2:11)
where W0 is the Hamiltonian of the atom in the absence of an external
field, m is the magnetic quantum number,33 L is the Larmor frequency
and g is the so-called Lande´ g factor. The g factors were introduced on a
purely empirical basis as Grundenergieniveau specific for each spectral
term.34 The magnetic quantum number m times g turned out to give the
right splitting factors for the anomalous doublets. For instance, by empir-
ically fixing the g values for the following spectral terms of the sodium
principal and sharp series doublets,35 the resultant mg splitting factors
could be obtained:
32
Landé (1921a).
33
Notice the difference between the first m in the formula (the magnetic quantum number),
and the second m in the denominator, which denotes the electron’s mass in the charge-
to-mass ratio of the Larmor frequency.
34
Landé (1921a). For a historical reconstruction, see Forman (1970).
35
Here the subscript of the spectral term denotes Landé’s half-integral j as well as the
corresponding magnetic number m, e.g. the term 2P1/2 corresponds to j ¼ 1/2 and
m ¼ 1/2. For details see White (1934), p. 158.
50 2 The origins of the exclusion principle
Landé proceeded to generalize this result to singlets and triplets. But how
to justify the g factors and their fractional values? How to explain the
2(2k 1) states observed in the presence of a magnetic field? Landé’s
phenomenological analysis left these questions unanswered. As we shall
see in the following sections, it was Heisenberg’s, Bohr’s, and Pauli’s task
to address them and find possible solutions.
Two years later, Landé returned to the problem.36 By using the atomic
core model he found an empirical formula for the g factors, which turned
out to depend on the quantum numbers j, r and k:
3 1 k2 r2
g ¼ 2 1 (2:12)
2 2 j 4
3 1 lðl þ 1Þ sðs þ 1Þ
g¼ (2:13)
2 2 j0 ðj0 þ 1Þ
where the pspin quantum number s has replaced the core quantum number
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
r and j ¼ jðj þ 1Þ. Landé’s procedure was entirely a posteriori: both the
g factors and the formula (2.12) were obtained from a phenomenological
analysis of spectra. Looking for a justification, Landé hypothesized – in
the light of the atomic core model – that the mg splitting factor corres-
ponded to the sum of the projections of the electron angular momentum
vector K and the core angular momentum vector R on the external field
vector H. Both K and R Larmor-precessed around the resultant total
angular momentum vector J, with J in turn Larmor-precessing around
H with angular velocity g!L. There was however a puzzling anomaly
concerning the observed unexpected value 2 of the g factor. In other
words, the mg splitting factor was equal to
rather than
36
Landé (1923a).
2.1 The prehistory of Pauli’s exclusion principle 51
37
In 1919 E. Beck performed a series of experiments that shed light on the existence of an
anomalous magneto-mechanical ratio for all ferromagnetic materials. However, Beck
thought that such an anomalous ratio was due to the electron’s orbital motion, and not to
the atomic core as later Landé (and after him Heisenberg) supposed.
38
An attempt to modify Larmor’s theorem without any compelling theoretical reason is
given in Landé (1921b), while the second option is discussed in Landé (1923a).
52 2 The origins of the exclusion principle
period between 1921 and 1924 Bohr, Heisenberg, and Pauli addressed
these questions and proposed alternative solutions that I now turn to in
some detail.
It follows from Landé’s analysis that the value of the component of the angular
momentum of the atom parallel to the field, divided by 2p, may not always be
equal to an integer, as assumed in the interpretation of the Zeeman effect of the
hydrogen lines . . . Landé actually obtained values for this quantity . . . differing
from an integer by 1/2. This finding has made questionable the justification of
fixing the motion in the stationary states of an atom with several electrons in
direct analogy with the quantum theory of periodic systems.42
39
Bohr (1923a). 40 Paschen and Back (1921).
41
In its simplest form, the Paschen–Back effect concerns the appearance of triplets that
resemble the Lorentz normal triplets of the normal Zeeman effect. In the terminology of
the semi-classical vector model, as the doublet splitting in the anomalous Zeeman effect is
due to the spin–orbit coupling, in which both k and s precess (with s twice as fast as k),
analogously the triplets of the Paschen–Back effect are due to the spin–orbit decoupling: in
a strong magnetic field, k and s get decoupled, separately quantized and precess around the
field independently of each other.
42
Bohr (1923a). Reprinted in Bohr (1977), p. 630.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 53
from the coupling of the core’s inner orbits with the valence electron’s
outer orbit. This resulted in radiative transitions not corresponding to the
fundamental harmonic component of the electronic motion, but to smaller
harmonic components: the correspondence principle still offered the gen-
eral framework for an understanding of complex spectra. Furthermore, he
noticed a ‘recurring closure of certain electron groups with increasing
atomic number, a closure expressing itself in a disinclination of these
groups to admit any additional electrons in orbits with the same values
of the quantum number n and k’.43 As we shall see in Section 2.2.4, what
Bohr here mildly describes as a ‘disinclination’ of groups to admit add-
itional electrons in orbits with the same quantum numbers soon turned out
to be a ‘prohibition’ according to Pauli’s exclusion rule. By contrast, Bohr
traced this ‘disinclination’ back to the symmetry requirements he envi-
saged for the closure of electronic orbits of noble gases,44 even if, as
mentioned in Section 2.1.1, these symmetry requirements implied a prob-
lematic reopening of already closed electronic orbits.
The interplay between inner and outer orbits (i.e. the core–electron
coupling) could – once suitably adjusted – account both for the alkali
doublets and for the anomalous Zeeman effect. In the case of alkalis,
Bohr fixed the j-values of the doublet p-terms (k ¼ 2) equal to k + d and
k + d – 1, whereas for the singlet s-term (k ¼ 1) j ¼ 1 + d, ‘where d is a
quantity that is independent of k, but is to begin with undetermined’.45
Bohr’s explanation of the anomalous Zeeman effect was even more ingenious.
As Landé’s analysis had revealed, an atom placed in a weak magnetic field
could take on 2(2k 1) states. But this number was not equal to the
product of the number of electron states with the number of core states
in the magnetic field. In fact, the atomic core could take on only one state,
i.e. perpendicular to the field direction, by analogy with noble gases. On
the other hand, the valence electron could take on only 2k positions in the
field so that the product of the respective positions of the core and electron
was 2k, a number half as large as Landé’s experimentally detected
2(2k 1). To accommodate this discrepancy, Bohr advanced the following
suggestion:
43
Ibid., p. 642.
44
‘We may assume that, in the normal states of the atoms of the noble gases, we are dealing
with electron configurations of pronounced symmetry. Actually, we see just in this
symmetry the explanation of the closed character of these electron groups; for the admission
of additional electrons, which would destroy this symmetry, would, in fact, not show that
analogy with a classical radiation process which could be established for the possible
transitions between stationary states of simply or multiply periodic systems.’ Ibid., p. 645.
45
Ibid., p. 645.
54 2 The origins of the exclusion principle
We are led to the view that, because of stability properties of the atom which
cannot be described mechanically, the coupling of the series electron to the
atomic core is subject to a constraint (Zwang) which is not analogous to the
effect of an external field, but which forces the atomic core to assume two
different positions in the atom, instead of the single orientation possible in a
constant external field, while at the same time, as a result of the same constraint,
the outer electron, instead of 2k possible orientations in an external field, can
only assume 2k –1 orientations in the atomic assemblage.46
Given the total angular momentum j of the atom, saying that its half-integral
values are due to the sum of k with a new quantity d ¼ 1/2, is empirically
equivalent to saying, in the later semi-classical terminology, that the total
46
Ibid., p. 646.
47
The adjective ‘non-mechanical’ was later added to Bohr’s ‘constraint’ by Heisenberg
(1925a). On Bohr’s unmechanischer Zwang, see Serwer (1977).
48
The hypothesis of a non-mechanical force not analogous to a Coulomb force had been
previously used by Van Vleck (1922) to explain the anomalous instability of the helium
model.
49
Bohr (1923a). Reprinted in Bohr (1977), p. 647.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 55
50
Ibid., footnote on p. 647. 51 Landé (1921a), (1921b).
52
Heisenberg (1922). For a detailed historical analysis of the first core model, see Cassidy
(1979) on which I draw for this section.
56 2 The origins of the exclusion principle
Z2p
p d ¼ k h (2:17)
0
H + HInt
HInt H
k* r*
γ
β
α
Fig. 2.4 The core–electron coupling in the presence of an external magnetic field
H, according to Heisenberg’s first core model. Adapted from Cassidy (1979).
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 57
53
In contrast with Landé, who identified the magnetic quantum number m with the
projection of the total angular momentum on H, Heisenberg identified m with the
projection of the electron angular momentum. Given the half-integral values for k*, m*
also took half-integral values (m* ¼ 1/2, 3/2, . . . , k) in agreement with Landé.
54
In September 1921 Sommerfeld (1922) was working on the Voigt theory of the anomalous
Zeeman effect for sodium D lines. Only with Heisenberg’s introduction of half-integral k
could the Sommerfeld–Voigt formula yield the optical doublets. See Cassidy (1979),
pp. 196–202.
58 2 The origins of the exclusion principle
l l
H H
S µs
µs S
Fig. 2.5 The spin–orbit coupling for the two fine-structure states (a) j ¼ l + 1/2
and (b) j ¼ l 1/2, according to the semi-classical spinning electron model.
Source: H. E. White (1934) Introduction to Atomic Spectra (New York,
London: McGraw-Hill Book Company). Reproduced with permission of The
McGraw-Hill Companies.
along the HInt + H direction, it can take on any continuous value. This
violated space quantization as well as the Rubinowicz rules: the total
angular momentum was not restricted to a set of discrete values. And
since Rubinowicz’s rules were obtained from the conservation of the
angular momentum in the classical theory of radiation, their violation
could be interpreted as due either to a violation of the conservation of
angular momentum or to a violation of the classical theory of radiation.
Between the two, Heisenberg sacrificed the latter by introducing non-
classical waves with fractional angular momenta, whose sum supposedly
yielded a classical wave with integral angular momentum.
This solution could not however prevent Heisenberg’s model from
facing even more theoretical difficulties, namely those Bohr was concerned
with. Fractional angular momenta violated the Bohr–Sommerfeld quan-
tum conditions. Furthermore, half-integral k were also a departure from
the relativistic X-ray doublets, for which the usual integral values were
used. Thus, Heisenberg’s first core model accounted for the optical doub-
lets in a way that contradicted the treatment of relativistic doublets: the
theoretical unification of optical and relativistic doublets was still far away
and the doublet riddle unsolved.
No less problematic was the half-integral value of the core. By picking
up half-quantum angular momentum from the valence electron, the
atomic core patently violated Bohr’s building-up principle. The quantum
numbers and the statistical weights did not remain adiabatically invariant.
The sharing principle strengthened the riddle of statistical weights.
To sum up, Heisenberg’s first core model could empirically entail alkali
doublets and their Zeeman splitting. But the price to pay for this empirical
success was the violation of several theoretical assumptions: Larmor’s
theorem, Rubinowicz’s selection rules, Sommerfeld’s space quantization,
the classical theory of radiation, and the Bohr–Sommerfeld quantum
conditions. Heisenberg announced his result in a letter to Pauli.
Intellectual honesty compelled him to admit the ‘drawbacks of this
theory’. Yet he optimistically concluded with the Machiavellian motto that
‘despite all: success justifies the means’.55 But, as a matter of fact, empirical
success does not justify the use of whatever means. No wonder then that
after two years Heisenberg proposed a second improved core model, which
in his expectations would succeed where the first model failed.
55
Heisenberg to Pauli, 19 November 1921. Pauli (1979), p. 44.
60 2 The origins of the exclusion principle
56 57
Heisenberg to Pauli, 19 February 1923. In Pauli (1979), p. 80. Landé (1923a).
58
Pauli to Landé, 23 September 1923. In Pauli (1979), p. 123.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 61
with the preceding noble gas because there were now two possibilities
for its angular momentum, either r or r + 1/2.
(ii) The Zweideutigkeit of the electron’s angular momentum shed light on
the problematic half-integral values for k, which turned out to be due
to k + 1/2, while integral-valued k was in agreement with X-ray
spectra.
Pauli’s notion of Zweideutigkeit exerted a direct influence on
Heisenberg. In May 1924, Heisenberg and Landé published a joint article
on the term-structure of higher-level multiplets,59 starting from Landé’s
recent analysis of the neon spectrum.60 The neon ion consisted of two
p-terms: a p1-term with j ¼ 2 and a p2-term with j ¼ 1. The addition of the
outer electron in the building-up process apparently made the p1-term split
into a triplet and a quintuplet, while the p2-term split into a singlet and a
triplet. Hence:
(i) from the neon ion ( j ¼ 1) with the addition of the valence electron, two
s-terms were derived with respectively j ¼ 1/2 and j ¼ 3/2;
(ii) from the neon ion ( j ¼ 2) another two s-terms were derived with j ¼ 3/2
and j ¼ 5/2.
Both these results seemed to spring from j ‘branching’ [Verzweigung] into
j 1/2 with the addition of the valence electron.61 Landé and Heisenberg
generalized this result in a ‘branching rule’ [Verzweigungsregel]:
When the atom is built up from the states of the ion that are characterized
through the j-values j1, j2, . . . , jn we assume that the atom possesses 2n s-terms
with the j-values j1 1/2, . . . , jn 1/2, and accordingly the atom shows 2n
multiplet-systems with multiplicity 2 j ¼ 2(j1 1/2), 2( j2 1/2), . . . and so on.62
This new rule, phenomenologically derived from the neon spectrum, could
account for the multiplet structure. Pauli’s Zweideutigkeit was tacitly at
work in the Heisenberg–Landé branching rule. The Zahlenpaar that Pauli
regarded as the ‘true’ angular momenta had here become the ‘branching’
quantum numbers of the total angular momentum. Yet the rule violated
Bohr’s building-up principle. Given the ‘branching’ process, the terms of the
atom could not be expected to coincide with those of the ion. The riddle of
statistical weights, once again, loomed on the horizon. Landé and
Heisenberg could not help concluding the article with an apologetic remark:
59
Landé and Heisenberg (1924). 60 Landé (1923c).
61
Landé and Heisenberg (1924), p. 280. 62 Ibid., p. 284.
62 2 The origins of the exclusion principle
63
Ibid., p. 286. 64 Heisenberg (1924). 65 See Serwer (1977), p. 217.
66
Heisenberg (1924), p. 301.
67
Heisenberg to Pauli, 7 December 1923. In Pauli (1979), p. 132.
68
The original title of Heisenberg’s work was ‘Über ein neues Quantenprinzip und dessen
Anwendung auf die Theorie der anomalen Zeemaneffekte’; a copy of the original
manuscript is in AHQP (45, 8).
69
Heisenberg (1924), p. 292.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 63
This rather obscure passage betrays the real nature of Heisenberg’s new
quantum principle. This was shaped along the lines of the recent
Bohr–Kramers–Slater73 (BKS) project of a quantum theory of radiation,
which was meant to be the highest expression of Bohr’s correspondence
70
Ibid., p. 293.
71
The Ersatzmodell was nothing but Landé’s atomic core vector model that he used for the
analysis of the anomalous Zeeman effect, as described in Section 2.1.3.
72
Heisenberg (1924), p. 299. 73 Bohr, Kramers, and Slater (1924).
64 2 The origins of the exclusion principle
74
Petruccioli (1988), English translation (1993), p. 120. See the same chapter for a detailed
analysis of the BKS theory.
75
Heisenberg to Pauli, 8 June 1924. In Pauli (1979), p. 155.
76
Heisenberg (1924), pp. 301, 303.
77
Heisenberg to Pauli, 21 February 1923. In Pauli (1979), p. 82.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 65
core.78 But he did not take the further crucial step of associating the
magnetic anomaly with the electron’s half-integral values. This step was
taken by Pauli, who dispensed with the atomic core and finally identified
the valence electron as the only culprit of all these anomalies.
78
Heisenberg (1924), p. 300.
79 80
Pauli to Kramers, 19 December 1923. In Pauli (1979), p. 135. Pauli (1923).
66 2 The origins of the exclusion principle
0, 1, . . . , (r 1/2) for odd multiplets (e.g. triplets). Following the program-
matic intent, no model-based interpretation was given for these quantum
numbers. Only in a later article (submitted in October 1923)81 did Pauli
explicitly fall back on the atomic core model in interpreting them as
m1 ! mk and ! mr
where the first is the orbital magnetic moment and the second the core
magnetic moment, so that the magnetic quantum number m ¼ m1 +
denoted the total magnetic moment of the atom. Without explicitly men-
tioning the atomic core model, in the first article Pauli went on to calculate
the Zeeman energy W ¼ W0 + mgh L (2.11) for strong magnetic fields
(where crucially g ¼ 1) as
W=hL ¼ m1 þ 2 ¼ m þ ¼ 2m m1 (2:22)
where once again the puzzling value 2 appeared in the Zeeman energy.
The strong-field term values for both even and odd multiplets followed.
Most interestingly, once these had been calculated, it was also possible
to get the term values for weak fields thanks to a new rule giving the
‘symmetry condition for the values of the terms in the transition from
strong to weak fields’,82 the so-called ‘permanence of the g sums’ as
Landé dubbed it:83
The sum of the energy values in all those stationary states belonging to given
values of m and k, remains a linear function of the field strength during an entire
transition from weak to strong field.84
81
Pauli (1924). 82 Pauli (1923), p. 156. 83 Landé (1923d).
84
Pauli (1923), p. 162. Notice that in the later semi-classical vector model for the spin–orbit
coupling, this rule came to mean that the sum of the g factors, for levels with the same total
angular momentum J or magnetic moment M, is the same in all field strengths,
independent of the coupling scheme used. In the semi-classical vector model, there are two
main coupling schemes for the spin s and the orbital angular momentum l of the valence
electron: the LS coupling (i.e. the so-called Russell–Saunders coupling) and the jj coupling.
In the LS coupling, given two valence electrons, the two l are coupled together to form the
resultant L and so also the two s to form the resultant S, where in turn L and S are coupled
together to give J. In the jj coupling, on the other hand, l1 is coupled with s1 to form j1 and l2
with s2 to form j2, where in turn j1 and j2 form the resultant J. The g factors change
depending on the LS coupling or jj coupling. Yet, according to Pauli’s permanence of the
g-sum rule, for given L, S, M or for given j1, j2, M the sum of the g factors remains the same
in the passage from strong to weak fields. See White (1934), pp. 189–91.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 67
And the search for an interpretation of the failure of Larmor’s theorem put
Pauli on the right track that neither Bohr nor Heisenberg had envisaged.
Pauli first tried to explain the failure of Larmor’s theorem by hypothesizing a
relativistic correction to the mass of the electrons in the core. In an important
letter to Landé,86 Pauli raised the question as to whether a deviation from the
85
Pauli to Bohr, 21 February 1924. In Pauli (1979), pp. 147–8.
86
Pauli to Landé, 10 November 1924. In Pauli (1979).
68 2 The origins of the exclusion principle
87
Ibid., p. 171.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 69
What would then follow in the case that the calculated deviations were not
effectively detected (as I think it is likely to be)? The entire calculation relies on
the assumption that the K-shell [i.e. the core] . . . possesses a non-vanishing
resultant angular momentum. But this momentum of the K-shell is not to be
taken too seriously and has something unrealistic about it . . . (I am tempted
to say that not only has the core momentum half-integral values, but also a
half-integral reality) . . . A future, adequate (not à la Heisenberg) theory of the
anomalous Zeeman effect must give reasons for the non-availability of the
relativistic correction factor.88
In the following few days Landé, from the Institute for Physics at
Tübingen, informed Pauli that the observed relativistic correction factor
in cadmium amounted to 6% and in zinc to 2%. But Pauli was
not quite satisfied with these results . . . One must consider elements with higher
atomic numbers . . . in order to decide the question with certainty . . . It is a pity
that the Zeeman effect is not precisely measured in alkalis with higher atomic
numbers. Do we know the precise values of the Ba spark spectrum? And those in
the III column? I am convinced since the beginning that there is no relativistic
effect.89
Landé finally announced the negative result for thallium (Z ¼ 90) that
Pauli welcomed as ‘totally sufficient to eliminate the last part of a possible
doubt about the empirical state of affairs’.90 The absence of a relativistic
correction in thallium spoke against the assumption of an atomic core
allegedly endowed with a non-zero angular momentum. Against this
assumption stood other facts that Pauli listed in the same important letter
to Landé on 24 November 1924. For instance, in the observable character-
istics, the K-shell did not differ from the closed shells with higher quantum
numbers. But the K-shell was typically ascribed a non-vanishing angular
momentum, whereas the closed shells with higher quantum numbers were
ascribed a zero angular momentum. This introduced an asymmetry in the
theory, to which nothing corresponded at the empirical level. Pauli drew
the following remarkable conclusion:
In the alkalis, the valence electron alone makes the complex structure as well as
the anomalous Zeeman effect. The contribution of the atomic core is out of
question (also in other elements). In a puzzling, non-mechanical way, the
valence electron manages to run about in two states with the same k but with
different angular momenta.91
88
Ibid., p. 171. 89 Pauli to Landé, 14 November 1924. In Pauli (1979), p. 172.
90
Pauli to Landé, 24 November 1924. In Pauli (1979), p. 176. 91 Ibid., p. 177.
70 2 The origins of the exclusion principle
92
Pauli (1925a), p. 385.
93
The Bohr–Coster theory of X-rays (1923) improved Sommerfeld’s earlier relativistic
treatment by identifying any X-ray term with three quantum numbers nk1, k2), with
k1 ¼ 1, and k2 ¼ 1,0. Landé (1923b) identified k2 with the inner quantum number j
of the alkali doublets, because it could take two states k2 ¼ k1, k1 1 as j did. In so doing he
took an important step towards the unification of relativistic and optical doublets.
2.2 Bohr, Heisenberg, and Pauli on spectroscopic anomalies 71
catch a glimpse of a logical connection between the number 2 of these states and
the half-integral quantum numbers, on the one side, and the violation of
Larmor’s theorem on the other side.94
From a free atomic core with Ni possible states springs a term-system with
(Ni + 1) (2k1 1) states and one with (Ni 1) (2k1 1) in the presence of a
field . . . Now I interpret these total 2Ni (2k1 1) states simply as follows: the Ni
states of the atomic core remain, and the valence electron takes 2 (2k 1) states
as in the alkalis.96
94
Pauli to Landé, 24 November 1924. In Pauli (1979), p. 178. 95 Ibid., p. 178.
96
Ibid., p. 178. 97 Ibid., p. 179.
98
Ibid., p. 180.
99
Stoner (1924). On Stoner’s work and its relevance to the origin of the exclusion principle
see the excellent article of Heilbron (1982), pp. 280–7.
72 2 The origins of the exclusion principle
There was still a long way to go for this prescriptive rule to become the
exclusion principle. As we shall see in Chapter 4, the process that led to
promoting this rule to the rank of a scientific principle is linked to the
development of quantum mechanics as a theoretical framework that
100
‘The strongest evidence for Stoner came from Alfred Fowler’s finding that the spectrum of
ionized carbon shows doublets that, because of their appearance and position, had to arise
from a 22 term. The natural interpretation of this unexpected oddity was, as Bohr himself
explained it, that ‘‘the singly charged carbon ion in its normal state besides two electrons in
21 orbits possesses one electron in a 22 orbit’’. That interpretation required a correction in
the interpretation he had laid down: the fifth electron for some reason does not take up the
21 path he [Bohr] had prepared for it. The reason, according to Stoner: since only two 21
orbits exist, the fifth electron necessarily falls into a 22 circle.’ Heilbron (1982), pp. 283–4.
101
On Rydberg’s formula, see Pauli (1955).
102
Pauli to Landé, 24 November 1924. In Pauli (1979), p. 180.
2.3 The turning point 73
Pauli’s rule came to be built into from the ground up. Hence, in this
original historical context, it is more appropriate to refer to it as a rule
than as a principle. Pauli himself called it Ausschlieungsregel103 or meine
Ausschluregel (exclusion rule),104 while Heisenberg teasingly called it
Pauli’s Verbot der äquivalenten Bahnen (Pauli’s prohibition of equivalent
orbits).105
Pauli admitted that ‘we cannot give a closer foundation to this rule, yet it
seems to present itself in a very natural way’.106 The exclusion rule was then
introduced as a theoretical consequence of reinterpreting Stoner’s rule in the
light of the electron’s Zweideutigkeit, with the help of the permanence of the
g-sum rule. The closure of electronic groups in the periodic table followed
naturally from it. Furthermore, the rule also accounted for the possible
combinations of electrons in non-closed shells. For instance, it finally
became clear why – despite alkaline earths containing two s-electrons in
their ground state, which could give rise to singlets and triplets – only
singlets were usually observed. The reason is that the two s-electrons are
equivalent and hence certain terms, namely triplets, are forbidden according
to Pauli’s rule. Only when one of the two electrons is excited to an s-orbit of
different n, does the triplet appear: similarly, for any two equivalent
p-electrons or d-electrons. Out of all logically possible combinations of
any two electrons, only a few of them can actually be realized: those in
which no two equivalent electrons are present. Everything seemed to fit
Pauli’s rule very well. Only Bohr’s correspondence principle was left out:
how to reconcile the classical periodic motions presupposed by the corres-
pondence principle with the classically non-describable Zweideutigkeit of
the electron’s angular momentum? Pauli could not help remarking that ‘this
is indeed a very strange state of affairs’.107
103
Pauli to Landé, 15 December 1924. In Pauli (1979), p. 191.
104
Pauli to Landé, 25 December 1924. In Pauli (1979), p. 196.
105
Heisenberg to Pauli, 16 November 1925. In Pauli (1979), p. 256.
106
Pauli (1925b), p. 776. 107 Pauli to Landé, 24 November 1924. In Pauli (1979), p. 180.
74 2 The origins of the exclusion principle
Dear Pauli,
I cannot easily describe how welcome your submission was. We are all excited
for the many new beautiful things you have brought to light. I do not need to
advance any criticism, since you by yourself, better than anyone else could have
done, have characterised the whole thing in your letter as complete madness . . .
I am not very sure either whether you are overstepping a dangerous threshold –
intoning your old ‘Carthaginem esse delendam’ – when you declare the definitive
death sentence about a correspondence-like explanation of the closure of
groups . . . I sense that we are here standing at a decisive turning point.108
108
Bohr to Pauli, 22 December 1924. In Pauli (1979), pp. 194–5.
109
‘Lead in its normal state has two superficial p-electrons (electrons with k ¼ 2). Five
possibilities therefore exist for the lowest state . . . There result five nkj terms defined by
values of j of 2, 2, 1, 0, 0, a prescription testable by calculating the g values and examining
the anomalous Zeeman effect in lead. Pauli was not pleased to find that only four levels
had been identified, and that they had been assigned j ’s of 1, 2, 1, 0. If the experimentalists
had not erred, he wrote Landé, ‘‘my closure rule must be modified for complicated cases’’
(Pauli to Landé, 15 and 25 December 1924). Heroic measures were called for. Pauli
stopped at the dull town of Tubingen en route to Hamburg from Vienna, where he had
spent Christmas. There, on Landé’s kitchen table, he examined Back’s latest photographs
of the lead-spectrum in a magnetic field. Analysis disclosed five terms with the predicted j
values.’ Heilbron (1982), p. 305.
2.3 The turning point 75
afternoon, still quite under the influence of the letter I had read, I succeeded in
deriving with it the so-called relativistic doublet formula.110
110
Kronig (1960), pp. 19–20. 111 Ibid., p. 21. 112
Pauli (1925b), p. 771.
113
Uhlenbeck and Goudsmit (1925).
76 2 The origins of the exclusion principle
114
Heisenberg to Pauli, 24 November 1925. In Pauli (1979), p. 265.
115
As Uhlenbeck later recalled, ‘It was quite clear that the picture of the rotating electron, if
taken seriously, would give rise to serious difficulties. For one thing, the magnetic energy
would be so large that by the equivalence of mass and energy the electron would have a
larger mass than the proton, or, if one sticks to the known mass, the electron would be
bigger than the whole atom! In any case, it seemed to be nonsense.’ Quotation from van
der Waerden (1960), p. 214.
116
‘Bohr’s optimism about Goudsmit’s theory has so much influenced me, that I’d really like
to believe in the magnetic electron.’ Heisenberg to Pauli, 24 December 1925. In Pauli
(1979), p. 271.
117
Bohr to Pauli, 20 February 1926. In Pauli (1979), p. 295. 118 Thomas (1926), (1927).
119
Pauli to Bohr, 26 February 1926. In Pauli (1979), p. 297.
120
Bohr to Pauli, 9 March 1926. In Pauli (1979), p. 309.
121
Pauli to Bohr, 12 March 1926. In Pauli (1979), p. 310.
122
Uhlenbeck and Goudsmit (1926). 123 Frenkel (1926).
2.3 The turning point 77
The Dirac relativistic wave equation for the electron finally allowed the
derivation of the spin magnetic moment, and in so doing vindicated Pauli’s
original reluctance to cast the electron’s Zweideutigkeit in the semi-classical
spinning electron model. But that is another story, and I shall come back to
it in Chapter 4.
124
Dirac (1928a), p. 610.