2번

VALENCE
and
The Structure of Atoms and Molecules
BY
GILBERT NEWTON LEWIS
PROFESSOR OF CHEM18TRY IN THE lfNIV£RSITY
OF CALIFORNIA
American Chemical Society

Monograph Series
BO O K DEPARTMENT
Th e CH EM I CAL CATALOG COMPANY, I nc.

'9 EAST '4TH STREET, NEW YORK, U. S. A.
19 2 3
COP\'RICHT, 1923, BY
The CHEMICAL CATALOG COMPANY, Inc.
All Rights Reserved
Pres'} of
J. J.Little & lves Company
New YOl'k, U. S. A.
GENERAL INTRODUCTION
American Chemical Society Series of

Scientific and Technologic Monographs
By arrangement with the Interallied Conference of Pure and
Applied Chemistry, which met in London and Brussels in July,
1919, the American Chemical Society was to undertake the pro-
duction and publication of Scientific and Technologic Mono-
graphs on chemical subjects. At the same time it was agreed
that the National Research Council, in cooperation with the
American Chemical Society and the American Physical Society,
should undertake the production and publication of Critical
Tables of Chemical and Physical Constants. The American
Chemical Society and the National Research Council mutually
agreed to care for these two fields of chemical development.
The American Chemical Society named as Trustees) to make
the necessary arrangements for the pUblication of the mono-
graphs, Charles L. Parsons, Secretary of the American Chemical
Society, Washington, D. C.; John E. Teeple, Treasurer of the
American Chemical Society, New York City; and Professor
Gellert Alleman of Swarthmore College. The Trustees have
arranged for the pUblication of the American Chemical Society
series of (a) Scientific and (b) Technologic Monographs by the
Chemical Catalog Company of New York City.
The Council, acting through the Committee on National Policy
of the American Chemical Society, appointed the editors, named
at the close of this introduction, to have charge of securillg
authors, and of considering critically the manuscripts prepared.
The editors of each series will endeavor to select topics which
are of current interest and authors who are recognized as author-
ities in their respective fields. The list of monographs thus far
secured appears in the publisher's own announcement elsewhere
in this volume.
4 GENERAL INTRODUCTION
The development of knowledge in all branches of science, and
especially in chemistry, ha. been so rapid during the last fifty
years and the fields covered by this development have been so
varied that it is difficult for any individual to keep in tonch with
the progress in branches of science outside his own specialty.
In spite of the facilities for the examination of the literature
given by Chemical Abstracts and such compendia as Beilstein's
Handbuch der Organischen Chemie, Richter's Lexikon, Ostwald's
Lehrbuch der Allgemeinen Chemie, Abegg's and Gmelin-Kraut's
Handbuch der Anorganischen Chemie and the English and
French Dictionaries of Chemistry, it often takes a grcat deal
of time to coordinate the knowledge available upon a single topic.
Consequently when men who have spent years in the study of
important subjects ure willlng to coordinate their knowledge
and present it in concise, readable form, they perform a service
of the highest value to their fellow chemists.
It was with a clear recognition of the usefulness of reviews of
this charucter that a Committee of the American Cllcmical
Society recommended the pUblication of the two series of mono-
graphs under the auspices of the Society.
Two rather distinct purposes are to be served by these mono-
graphs. The first purpose, whose fulfilment will probably render
to chemists in general the most important service, is to present
the knowledge available upon the chosen topic in a readable
form, intelligible to those whose activities may be along a wholly
dIfferent line. Many chemists fail to realize how closely their
investigations may be connected with other work which on the
surface appears far afield from their own. These monographs
will enable such men to form closer contact with tile work of
chemists in other lines of research. The second purpose is to
promote research in the branch of science covered by the mono-
graph, by furnishing a well digested survey of the progress
already made in that field and by pointing out directions in
which investigation needs to be extended. To facilitate the
attaimnent of this purpose, it is intended to include extended
references to the literature, whieh will enable anyone interested
to lollow up the subject in more detail. If the literature is so
voluminous that a complete bibliography is impracticable, a
critical selection will be made of those papers which are most
important.
GENERAL INTRODUCTION 5
The pUblication of these books marks a distinct departure in
the policy of the American Chemical Society inasmuch as it is
a serious attempt to found an American themical literature with..
out primary regard to commercial considerations. The success
of the venture will depend in large part upon the measure of
cooperation which cnn be secured in the preparation of books
dealing adequately with topics of genera] interest; it is earnestly
hoped, therefore, thnt every member of the various organizations
in the chemical and allied industries will recognize the impor-
tanre of the enterprise and take sufficient interest to justify it.
AMERICAN CHEMICAL SOCIETY
BOARD OF EDITORS
Scientific Series:- Technologic Series:-

WILLIAM A. NOYES, Editor, HARRISON E. HOWE, Editor,
GILBERT N. LEWIS, C. G. DERICK,
LAFAYETTE B. MENDEL, F. A. L,DBURY,
ARTHUR A. NoyES, ARTHUR V. LITl'LE,
JULIUS STIEGLITZ. C. P. TOWNSEND,
JOHN JOHNSTON,
R. E. "\\'ILSON.
American Chemical Society
MONOGRAPH SERIES
Other monographs in the series of which this book is a part
now ready or in process of being printed or written.
Organic Compounds of Mercury.
By Frank C. Whitmore. 397 pages. Price $4.50.
Industrial llydrogcn. By Hugh S. 'Taylor. Price S3})O.
The Chemwtry oj Enzyme Actions.
By K. George Falk. 140 pages, Price $2.50.
The Vitamim.
By H. C. Sherman and S. L. Smith. 273 pages, Price $4.00.
T'By c;~~~~ilcf{i~td. O{861~~~t.ar~~f: S~O~lectro1l$'
Zirconium and Its Compounds.
By F. P. Venable. Price $2.50.
The Properties of Electrically Conducting Syste-rns,
By Charles A. Kraus. Price $4.50.
The Analysis of Rubber. By .John R Tuttle. Price $2 ..50.
The Origin of Spectra.
By Paul D. Foote and F. L. Mohler. Price $4.50.
Carotinoids and Related Pigments.
By Leroy S. Palmer. Price $4.50.
Glu.c (lnd Gelatin" By .JCI"OIIlC Alexander. Pricf' $3.00.
The Chemistry of Leather Manufacture.
By John A. Wilson. Price $5.00.
Wood Distillation. By L. F. Hawley. Price $3.00.
Thyroxin. By E. C. KendalL
The Properties of Silica and the Silicates. By Robert B. Sosman.
Coal Carhonizat.ion. By Horace C. Porter.
The Corrosion of Alloys. By C. G. Fink.
Piezo-Chcmwtry. By L. H. Adams.
Cyanamide. By Joseph M. Braham.
Liquid Ammonia as a Sollient. By E. C. Franklin.
Solubility. By Joel H. Hildebrand.
Organic Arsenical Compounds. By George W. Raizlss. Jo-
seph L. GavTon.
Shale Oil. By Ralpb H. McKee.
Alumino_thcrmic Reduction of Metals. By B. D. Saklatwalla.
Absorpttve Carbon. By N. K. Chaney.
Refining of Petrole-u:'m. }3y George A. Burrell. et af.
Extraction of Gasohne from Natural Gas. By George A. Burrell.
The Animal as a Concerter. By H. P. Armsby and C. Robert
Moulton.
Chemistry of Cellulose. By Harold Hibbert.
The Properties of Metallic Substances. By Charles A. Kraus.
The Structure oj Crystals. By Ralph W. G. Wyckoff.
PhDto.~ynthesis. By H. A. Spoehr.
Colloid Chemistry. By The Svedberg.
Phyaicai and Chemical Properties of Glass. By George W.
Morey.
The Chemistry of the Treatment of Water and Sewage. By
A. M. Buswell.
The Chemistry of Wheat Flour. By C. H. Ba.iley.
The Rare Gases ollhe Atmosphere. By Richard R Moore.
The Manufacture of Sulfuric Acid. By Andrew M. Fairlie.
l'he Chemical Aspects oj Immunity. By H. Gideon Wells.
Protective MelaUic Coatings. By Henry S. Rawdon,
The CHEMICAL CATALOG COMPANY, Inc.

19 EAST 24TH STREET, NEW YORK, U. S. A.
To my colleagues and students of the University of
California, without whose help this book would not
have been written. In our many years of discussion
of the problems of atomic and molecular strncture,
some of the ideas here presented have sprung from
the group rather than from an individual; S0 that in
a senst J am acting only as editor for thi" group.
PREFACE
I take it that a monograph of this sort helongs to the ephemeral
literature of science. The studied cafe which is warranted in the
treatment of the more slowly mO\'ing brandH..:s of science would he (lut
of place here. Hather with the pen of a journalist we l11u:;t attempt
to record a momentary phase of current thought, which may at any
jn~tant chan~e with kaleidoscopic ahruptness.
It is therefore not unlikely that some of the things said in this hook
may soon have to he unsaid, hut I trust that these may be mattr-fS of
detail rather than of essence. During the s~'v('n years that han' elapsed
sinn' my prcyiol1s publication concerning the structUT(_~ of the molecule
and the nature of the chemical bond, I have found little need of suh-
tracting frotTI the views there set forth. although there i:-> now much to
add. :-:::'0 in this present work 1 shall hope that there are no serious ~ins
of commission. That there arc sins of omis:-;iol1 r am <lln'adv only tot)
well aware. To attempt to keep pace with the rapid d{'\Tlopn;ents 'in so
many ramifIcations of science, all of which rontrilmte 10 our knowl-
ed~c of the atom and the molecule, is. especially for one who is at hest
a slothful reader, an impossible task.
K e\'erthe1ess it is the same atom and the same molecule that is heing
studied hy the organic chemist. the inorganic chemist and the physicist;
the marvellously exact conclusions of the spectroscopist, the far more
vague hut equally diffil'tl1t and important generalizations of the student
of the carhon compounds. must ('ontrihute, each in due measure, to
our knowledge of that microcosmos which appears to us the more
mysterious as its nature becomes more nearly reycalerl tn us. It was
with this thought in mind that I have devoted several of the earlier
chapters to an attempt 10 bring to the hetter acquaintance of chemists
some of the astounding a('c(lmplish111ent~ of modern Ilhysics.
GILBERT N. LEWIS
CONTENTS
CHAPTER 1. THE ATOMIC THEORY 17
THE DISCONTINUITY OF lVfATTER. 17. The ,Vark of Dalton,

17· THE UNITY OF 1\1ATTER, 18. Prout's Hypothesis and
Its Recent Revival. 18. THEORIES OF CHEMICAL AFFINITY,
20. The Electrochemical or Dualistic J Jypothcsis, 20. Its
Conflict with the Theories of Structural Organic Chemistry,
21. Faraday's Law and the Discontinuous Nature of Elec-
tricity,21. The Theory of Arrhenius, 21. Is there an Essen-
tial Distinction Letween "Polar!! and "N onpJlar" Com-
pounds? 22.
CHAPTER II. TIlE PERlOllIC L\w .\.:\1) THE CIIE:\fIST'S PiCTURE

OF THE ATOM
The Development of the Periodic Tahle, 23. Ryrlberg's Or-
dinal NumLers, 24. Tahle of Atomic N"urnhers and the \Vork
of l\loseley, 25. A Convenient Form of Periodic Table, 28.
SOME ATOMIC MOJ)ELS, 29, Ahegg's Normal and Contra-
Valences, 30. Thomson's Model of the Atom, 31. Ruther-
ford's Planetary Theory, 32. Parson's l'dagnctun Theory,
32. Theories of Kossel and Lewis, 33.
CHAPTER III. SPECTRAL SERIES A::\'1J THE PIIYSTCIST'S VlE\,·; OF

THE ATOM 35
The Classical Theory of the Emisslon and Absorption of
Llght, 35. The Balmer Formula, 36. The Pickering Series,
37. THE WORK OF RYDBERG. 37. The Rydberg COflstant.
37. The Combination Principle, 37. A Spectral Line as a
Difference of Two Frequencies, 38. THE QUANTUM THEORY,
39. Departure from the Equipartition Law, 40. Planck's
Hypothesis, 41. The Einstein Equation, 42. A PARTIAL
STATEMENT OF BOI-IR'S THEORY, 43. Energy Levels, 43.
THE INTERPRETATION OF X-RAY SPECTRA, 45. IONIZATION
AND RESONANCE POTENTIALS, 46. BOHR'S ATOMIC MODEL,
47. Quantum Orbits, 48. Calculation of the Rydberg Con-
stant, 49. The Pickering Series as a Helium Spectrum, 49·
Certain Possible Objections to Bohr's Hypothesis • .10. The
Importance of the Electron Orbit as a Unitary Entity, 51.
MAGNETIC PHENOMENA. 52. :Magnetic ~iotnent, 53. Para-
and Diamagnetism, 53·
II
I2 TABLE OF CONTENTS
ClIAPTBR IV. REC01\~CILLI\TION OF THE Two VIEWS; THE AR-

RIINGEME;'\'T UF EJ.ECT1WKS IX THE ATO)f 55
The Three-Dimensional Character of Atomic Structure. 55.
The Electron Orbit as the Building-Stone of Atomic Struc-
ture, 56. Summary of Existing Information. 56. The Im-
portance of the Group of Eight, 57. The Still Greater Im-
portance of the Group of Two, Of the Electron Pair. 57.
Bohr'~ \~iews Regarding the l\1aximum Xumher of Electrons
in Certain Groups and Suh-Groups, SR. TIlE 11\ XER STRCC-
TURE O.F TIlE SEVERAL AT(nr~. 5X. Certain Orhital Arrange-
ments Inco11l}k1.tihle with ()!Jseryed :rvlagnetic Phenomena, 59.
Two \\'ays in which a Pair of Elementary 11agncts lIlay Be
Coupled, 59. The Elements of the Short Periods. S9. THE
FIRST LOK{; PERInI), ()O. Tautomerism within the Kernel, or.
Composition of the ,\tnmic Kernels in Various Elementary
1(HlS, 62. The 1)aramagneti~m of Elements with Uncom-
pleted Kernels, (q_. TIlE RE!>.-IAI,:\j]'\G PERIODS, (\s. Two
Important Phenomena Still Unexplained by Quantum
Theory, 66.
CIL<\J'TER V. TllE t~]'OO:.";' OF A1'OMS; TIlE MODERN DLTALISTIC

THEORY • 67
Failure of the Older Valence Theon', 68. The \Vork of
'Verner, GR Earlier Amhiguities in the Term "Valenre," 68.
The Modern Dualistic Theory and Its Inadequacy. 71. The
De\Tlopment of the Tdea that Two Atoms 11ay Hold Elec-
trons in Common, i4.
CILA,I'TER Y1. THE l\~E\y TllEORY OF V.\LENCE; THE CHEMICAL

BO]\'D • 79
Satisfaction of Atomi{_' Octets by the Sharing of Electrons,
79. THE PA1RI);"\:" OF ELECTRONS. 79. The Rarity of Odd
),jolecules and Their Peculiar Properties, So. The Physical
Importance of the Phenomenon of Pairing, 81. THE BOND,
8T. The Chemical nond is Always a Pair of Electrons
which Lies between Two Atomic Centers and is Held Jointly
in the Shells of the Two Atoms, 81. The Tetrahedral Char-
acter of the Normal Octet. 82. The Essential Identitv of
the Bond in Polar and Nonpolar Compounds, 83. Electrical
Polarization Due to the Shifting of Electron Pairs, 84. The
New Theory Includes the Modern Dualistic Theorv as an
Extrel'tle Case, 86. OTHER FEATVRES OF THE NEW VALENCE
THEORY, 86. Singly Bonded Oxygen. 86. The Universality
of a Few Simple Structural Types, 86. The Quadrivalence
of Nitrogen, 87.
TABLE OF CONTENTS '3
CHAPTER VII. DOUBLE AND TRIPLE BONDS • 88

The Unsaturated Character of the Douhle Bond. 8R The
1fagnetic ProlJerties of the Douhle Bond, 89. The Haeyer
Strain Theory, &J. Conjt1Rated Double Honds. qo. The
"Partial Valence" of Thiele. <)l. Huggins' Theory of Con-
jugation, 91. The Structure of Benzene, (p. TIlE TRIPLE
BOND. 93. The Comparatively Greater Saturation of the
Triple Bond, 93. To \Vhat Extent is the Triple Bond a
H.ealityr 93. LIMlTATlOK OF l\1PLTIPLE HO]'\f)s. t)-t.. ~'lul
tiple Bonds Limited to Elements of the First J leriod of
Eight, 94. Eastman's Theory, <)5· Rl~~l)1ri·:. 96.
CUAPTER VIII. EXCEPTlOl"S TO TIlE RULE OF EIGHT 97

Case:;; in which the ,\tomir ~hcll Contains Few('r than Four
Electron Pairs, 9/. The Boron Compounds. 9 g . Sulfur
Trioxide and Similar Compounds, 99. ATOMS WITH T\tORE
THAI" FOUR ELECTRON PAIRS, lor.
CHAPTER IX. 'VALENCE AND COORDI'K/\TlO!I.~ Nl1MHER 104
A Redefinition of Valence. J04. The Normal Valence of

Four, 104. "Onium" Compounds, 10;;. A Common Type
of Tautomerism. lOR. BIVALEKT HYDlWGEN, lOC). The
Hydrogen Bond. W(). The \Vork of Latimer and H.ode1Jt1~h.
lOt). TIlE ~l'ADRl\'ALE;';CE OF NLTROC;EN, 1]1. The Amine
Oxides and Their Compounds, Ill. Do C:01111x1unds of
Quinqui1-alent ~itrogen Exist: ] I3. The \Valden Im'crsinn,
I.l3. VALENCEt-i Hl(;llER 'fILA1\: FOt:R, 113. j\re "Valence"
and "Coordination Numher" Synonymo11s~, ] IS. VALENCE
IN COI\'DENSEl> SYSTEMS. IT6. The Stmcture of SoJids
and Liquids, 116. Rf:Sl'ME, 118.
CHAPTER X. COMPOUNDS OF ELEMENTS \YITH SMALL KERNELS II9
Molecular Radii, II9. HYDROGEN AND IIELICM, J2J.~

LITHle1f AKD BERY1.LICM, 122. BORON, 123. The Unique
Character uf the Hydtoborons, 123_ NlTROGEN AND CAR-
BON, 124. A Further Discussion of Multiple Bonds, 124.
The Essential Identity in Structure of Nitrogen, Carbon
Monoxide, Cyanide Ion and Double Acetylide Ion, 127.
Peculiarities of Nitric Oxide and the Nitroso-Compounds,
128. OXYGEN AND FLUORINE, 129. A New Formula for
Ozone, 130. Possible Compounds between Fluorine and
Oxygen, 130,
'4 TABLE OF CONTENTS
CHAPTER XI. ELEMENTS IN POSITIVE AND NEGATIVE STATES 131
Criticism of our Present Terminology, 131. THE PROBLEM

OF ELECTROMERS, 131. COMPOUNDS BETWEEN NEGATIVE
ELEMENTS, 132, The Bond in a State of Strain, 133. The
Problem of "Positive Chlorine," 133. The Localization of
Electric Charge within the Molecule, 133. The Basic Prop-
erties of an Element in a Negative State, 134. A Possible
Mechanism of the Decomposition of Alkoxylamine, 135-
Aldehyde-Ammonia, 136.
CHAPTER XlI. REMNANTS OF THE ELECTROCiIEl\ftCAL TUEORY 137

The Inflw:nce of an Electric Charge upon Further Ionizatlon,
137. TnE STRENGTH OF ACIDS AND BASE..'-;, 138. The In-
fluence of a Negatiyc Substituent upon the Strenh>th of an
Acid, 139. The Effect Communicated through a Chain, 139,
The Amino-Acids, 139. The Dibasic Acids, 139. Certain
Conclusions Regarding the Structure of Inorganic Adds,
140. TllE DEF1K1TION OF AnDS AKD BASES, 141. The
Definition of an Acid or Dase in a Given Solvent, 142, A
110re General Definition of EssentiaIly Acid and nasic Sub-
stances, 142. OTHER FACTORS DETERMINING DISSOCIATION,
'42. The Effect of Mobility in the Electron Structure, 143.
The Effect of Spatial Position; Cis- and Trans-~\cids, 144.
THE RULE OF CRGM DROVIN AND G1BSO!', 144. Argu-
ments again:;,t a rrononnced Alternation in Electrical Polar-
ity, ]45. Alternatlon of Residual Affinity,' 145 .. rossible
Polarization Just Preceding a Reaction, 146. I{tsu"UE, 146.
CUAPTER XIII. THE SOURCE OF CHEMICAL AFFINITY; A 1\1AG-

NETOCBEMlcAL THEORY 147
Simple Electrostatic Forces Incompetent to Account for the
1\lost Essential Chemical Phenomena, 147. The Conjuga-
tion ,Of Elementary 'Mag:nets, 147. The Elimination of
Magnetic Moment, '47. The Odd Molecule Represents the
Highest Degree of Electrical and :Magnetic Ul1saturation,
148. The Neutralization of Magnetic Fields by the Coupling
of Electrons, '49. The Further Elimination of the Residual
Magnetic Fields by the Formation of Octets, '49. Distor-
tion of the Stable Magnetic Structure Increases the Mobility
of Electrons, r50. Is the Stablest Structure One in which
Distortion is Equally Distributed through a Molecule? ISO.
Chemical and Magnetic Unsaturation Due to Double Bonds,
151. Diamagnetism and Paramagnetism, 152. Is the Mag-
netic Moment Always a Multiple of a Quantum Unit? '52.
TABLE OF CO.\'TENTS IS
CONJl'(;ATlON. ]53. Chemical Conjt1~ati()11 h l\ta~n('tit

Conjugation, J53. Benzene. ]54. The Carboxyl Group,
154. Hal{)gen l~onds, 156. Diazo-Compounds, 156.
CUAPTER XIV. THE DISCONTINTITY OF PHYSICo-CHEMICAL

PROCESSES 157
The Photochemical Themy of Hcactiol1S, 157. Its Inadequacy
to Account for All Hc:aniotls. 157. THE DtSCO:STINUITV
OF ell E:>'11CAL PROCESSES, 1 SR. Resonam'(' and Ionization as
the Sjll1ple~t Chemical I\.('a~·ti(ln~, J SR Discontinuity in the
~Yl1tl1f'sis of Ammonia, J.1R The Meaning of "Tight" and
"Loose" nomb in Terms of the Quantum Theory. 159. The
Hupture (If a Loose l~(lnd. 160. Catalysis, 160. COLOR, 160.
The Translation of the Older Color Theory into the L.1.11-
gunge of Quantum Theory. 16r. Color 85 F.vidence (If the
Nearlless of Energy Levels, 161. Thc~ Mobility of Elec-
trons in Gases, 162, THE Fl'Tl'RE OIT 'QUAKTCM THEORY,
J(I::;. The Nature of 8 Field of Forc('. 16-+. The Nature of
Space, 16.+. A Network GeometrY, 105.
HEFEHENCES 166
ff,}~~~
r;;O!( \, ~H'I\O?"
~~" ~
VALENCE AND THE STRUCTi"T'D''iii;;;;;;iiiiii"-

~
ATOMS AND MOLECULES.
Chapter 1.
The Atomic Theory.
The Discontinuity of Matter.
There has been much debate among historians of chemistry as to
the order of discovery of Dalton's two great generalizations. Which
came first, the atomic theory or the law of multiple proportions? The
fact probably is that in Dalton's mind the two ideas were essentially
one. The concept of a granular structure of matter had been a favorite
among philosophers for centuries, and at the beginning of the nineteenth
century it was prevalent among scientists and laymen. "Pound St.
Paul's church into atoms, and consider any atom . . ." Boswell quoted
Johnson as saying a decade earlier.
Moreover, many of those who held this philosophical doctrine
regarded the atoms of anyone simple substance as equivalent to one
another, like building bricks. The idea that a simple substance is com-
posed of small particles. all similar one to another, must therefore be
presumed to have been a part of the intellectual heritage of that period.
What Dalton saw (1808) was the possibility of a crucial scientific
test of this hypothesis. If elements and compounds are made np of
discrete and characteristic particles, each particle of a compound sub~
stance must contain an integral number of particles of its component
elements. In fact Dalton fonnd, in the two hydrocarbons which we
call ethylene and methane, a given amount of hydrogen combined with
twice as much carbon in the former as in the latter. In the two oxides
of carbon he found the ratio of oxygen to carbon to be twice as great
in one as in the otber. When he also discovered a similar integral
:eJationship among the oxides of nitrogen, he felt justified in announc-
Ittg the general law of multiple proportions. The crudity of the experi-
ments upon which he hased this law, and the fact that his analysis of
one of the oxides of nitrogen was entirely erroneous, indicate a strong
predisposition toward the conclusion which he reached.
The law of multiple proportions converted a philosophic speculation
into a working theory of science. The theory of atoms and molecules
17
18 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
not only became the basis of stoichiometry, but later again proved its
great fertility in the development of the mechanical theory of heat.
At the end of the last century a short period of scepticism as to the
reality of atoms and molecules was abruptly terminated through the
rapid advance of scientific discovery. It became possible to count the
atoms. The ultramicroscope in the hands of Perrin (lgoB) permitted
the observation of particles moving about in exact accord with the
predictions of molecular theory. In the processes of radioactivity heavy
atoms were found to be disintegrating into lighter atoms, and it was
shown by Rutherford and Soddy (1903) that each atom of helium
emitted by a radioactive substance produces a scintillation upon a screen
of barium platino-cyanide, so that in a sense we see the individual
atoms.
Thus the atoms have been counted. analyzed. decomposed. Even
the secrets of the innermost nucleus are being brought to light. But
such familiarity has tended to produce over-confidence. For a time it
seemed that the structure and the behavior of the atom could be inter-
preted without essential change in the modes of thought which had
been found adequate in dealing with the massive bodies of everyday
experience. However, this feeling of confidence has received a rude
shock as we have encountered one by one the mysteries and paradoxes
which have led to the quantum theory of the present day.
Through the work ·of Dalton the conception of matter as a con-
tinuum was definitely displaced by the conception of discrete quanta
of matter J and we are now beginning to see that this was but the first
stage in a great revolution against the theory of the continuum. Step
by step we are being forced to "quantize" physico-chemical phenomena.
How far this revolution will go, and how much of our former belief
in the continuity of nature will remain, we cannot now predict; but
it is already evident that many of our best established principles of
science are under fire, and we may be sure that the theory of atoms is
but one of many phases of the coming theory of discontinuity in nature.
The Unity of Matter.

There is another philosophic helief which at all times has been
widely held. This is the idea that all of the various substances known
to us are merely different manifestations of a single basic substance.
Just as Dahon saw the scientific implications of the atomic theory, SO
Prout (1815) saw a possible scientific mnsequeuce of the theory of the
unity of matter.
He noted that the weights of the several atoms appeared to be
multiples of the weight of hydrogen, and advanced the idea that all
other atoms are composed of hydrogen atoms. This proposal, which
was vigorously contested, received the adherence of some of the best
minds of the period. The experimental evidence waS conflicting.
Atomic weights had been but roughly determined, and while accidental
errors would not on the average bring atomic weights nearer to whote
THE ATOMIC THEORY 19
numbers, an instinctive tendency toward the rounding off of uncertain

figures seemed adequate to account for the rule discovered by Prout.
~ontinual refinement of method led to an increasing accuracy in
analytical results, and for over a century the determination of atomic
·w'eights has been one of the favorite occupations of chemists. I t soon
became apparent that the atomic weights in general were 110t exact
multiples of that of hydrogen, and Prout's theory gradually fell into
disrepute. Nevertheless it was occasionally pointe<.i out th.at the
majority of atomic weights were much nearer to whole numbers than
would be expected from tf~e laws of chance.
For example. Rydberg (1897) showed that the chance of the
atomic weights of the first twenty-two elements falling as near to
whole numbers as they do would be less than one in one billion. It
therefore seemed reasonable to ascribe the close approximation of atomic
weights to whole numbers, not to pure chance, hut rather to some such
fundamental principle as that of Prout, perhaps modified by some
factors of a secondary nature. Indeed we are now nearly convinced
that Prout's theory was correct, and that the deviations of atomic
weights from integral numbers are due to two separate causes.
The first of these causes was foreseen in a remarkable prophecy
made by Marignac in r860. He says: "Could one not, for example,
while preserving the fundamental principle of this law (of Prout)
make the following supposition, to which I do not attach importance
except in s9 far as to show that one might explain· the dis(.ordance
. which appears to exist between the results of observation and the
immediate consequences of this principle? Might one not suppose that
the cause, which is unknown but probably differs from the physical
anti chemical agencies that we recognize, and which has uetermined
certain groupings of the atoms of the single primordial matter tQ give
birth to our simp-Ie chemical atoms, and to impose upon each of these
grou ps a special character and particular properties, has also been able
to exercise an influence upon the way in which these groups of atoms
obey the law of universal attraction, such that the weight of each of
them is not exactly the sum of the weights of the primordial atoms
which constitute them?" Since the advent of relativity we know that
the mass of a body varies with its energy, so that if two atoms combine
with a certain evolution of energy there is a proportional loss in mass.
This is one of the reasons for the deviations from the rule of Prout.
The second cause which we now recognize as responsible for some
of the large deviations from the rule of Prout is that many of the
elements are mixtures of one or more isotopes which can be sepa-
rated only with the greatest difficulty. Such elements have atomic
weights that depend upon the relative amounts of the several isotopes
which they contain. The separate isotopes, studied by the method of
positive ray analysis developed by J. }. Thomson (1913) and by
Aston ('920), for the most part show atomic weights which are very
dose to whole numbers.
The probleni of is9t<>pes is one which concerns what is now called
.. VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
the nucleus of the atom, and it would carry us too far from the main
purpose of this work to discuss in any detail the many important
observations which concern the structure and the disintegration of that
nudeus.
Theories of Chemical Affinity.
In the earliest experiments on electricity it was found that different
substances, brought into contact and then separated, remain charged,
oue with vitreous and the other with resinous electricity, or in Frank-
lin's nomenclature, with positive and negative electricity. Different
substances appeared to exercise different degrees of attraction for the
electric fluid or fluids.
In the course of the brilliant experiments of Davy (r&>7) concern-
ing the effect of the electric current upon various chemicals, he was
led to the idea that the particles of substances become electrified when
they meet dissimilar particles, and that the attraction between the oppo-
site charges, so produced, is the cause of chemical union. This idea
was developed by Berzelius (r8r9) into the electrochemical theory,
which remained for many years the dominant theory of chemistry.
Al1 kinds of chemical union were explained in accordance with
this theory. Thus it was considered, when an atom of zinc comes in
contact with an atom of oxygen, that a flow of electricity occurs which
leaves the former positive and the latter negative. Although sulfur
also would be negative toward zinc it is positive toward oxygen and
becomes the positive part of a molecule such as sulfur trioxide. Thus.
each of the molecules, zinc oxide and sulfur trioxide, was regarded
as held together by the electric forces operating between the oppositely
charged parts. But these two molecules when brought together would
also not remain neutral. Zinc oxide as a whole being positive with
respect to sulfur trioxide, these two molecules would in turn be held
together by electric forces to produce a molecule of zinc sulfate. Soon
the theory was extended to apply not only to simple compounds but
even to the most complex bodies known to mineralogy.
When the electrochemical or dualistic theory was first proposed it
was not known that some of the firmest chemical compounds are com-
posed of two like atoms, as H, or N ,. The existence of such types of
union presented an apparently insuperable objection to the theory.
Also the study of organic chemistry drew attention to a class of com-
pounds which seemed to fit inadequately into the dualistic scheme of
Berzelius. Especially it was pointed out that electronegative chlorine
could be substituted for electropositive hydrogen, in numerous com-
pounds, without appearing to produce any pronounced change in prop-
erties. In consequence of these discoveries the dualistic theory was
largely abandoned. ,
Then continued the great development of structural organic cbem-
istry, from the work of Kekule (r858) on the chemical bond and on the
spatial arrangement of the atoms, to tbe work,of LeBel (r875) and
of van't Hoff (r875) on stereoisomerism. No generalization of science,
THE ATOMIC THEORY 21
even if we include those capable of exact mathematical statement, has

ever achieved a greater success in assembling in simple fonn a multitude
of heterogeneous observations than this group of ideas which we caIl
structural theory. The graphical formula is far more than a mere
theory of atomic arrangements; it has become a' remarkable shorthand
method of representing a great variety of chemical knowledge.
In all this development of structural chemistry the electrochemical
properties of the elements seenled to play but a subordinate role. But,
after the great wave of enthusiasm for the synthesis and analysis of
complex organic substances was spent, attention once more reverted to
substances of the saline type. Faraday (1833) had shown that the
law of definite and multiple proportions is valid not only for the
chemical elements but also for electricity. Thus a gram of copper
carries just twice as much electricity in the electrolysis of a cupric salt
as in the electrolysis of a cuprous salt. It is singular that so many
years elapsed after the announcement of Faraday's law before it was
realized that this law implies a discontinuity of electricity in the same
degree that Dalton's law implied a discontinuous structure of ordinary
matter. By the same reasoning electricity occurs in quanta that are
all alike, and capable of combining with atoms and groups of atoms only
by integral numbers. It was Helmholtz in his celebrated Faraday
lecture of 1881 who first pointed out this deduction of the atom of
electricity, or as it is now called, the electron.
Our knowledge of the atom of negative e1ectricity, the electron, is
largely due to the brilliant investigations of J. J. Thomson and of those
whom he has inspired. The proof that free electricity is negative elec-
tricity; the determination of the ratio between the charge and the mass
of an electron; and the study of the physical and chemical effects pro-
duced by moving electrons, comprise one of the most fascinating chap-
ters of modern science.
The study of saline substances was greatly fostered by the electro-
lytic dissociation theory of Arrhenius (1887) which clarified in so
remarkable a manner our ideas concerning salt so1utions. This theory
through a generation of criticism has fully justified its essential
accuracy. We are fully convinced that, in a dilute aqueous solution of
sodium chloride, this salt is separated into two distinct parts, one of
which has a negative charge, equal to the charge of an electron, while
the other is positively charged in equal amount. Thus we have full
demonstration of a phenomenon which was assumed in the dualistic,
theory.
Again chemists were tempted to revert to the electrochet11ical theory
as an explanation of all chemical union, and again they met the diffi-
culty of explaining by such means the properties of substances like
methane and diatomic hydrogen. There obviously is a wide gap between
extreme types: on the one hand an extremely "polar" substance like
sodium chloride, in which presumably there is at all times a considerable
displacement of electricity from the sodium to the chlorine, and which
sometimes completely dissociates into sodium and chloride ions; on
:n VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
the other hand a relatively non-polar substance like diatomic hydrogen

which gives no a priori reason for, nor shows any evidence of, such
electric displacement. Must we conclude that there are two distinct
types of chemical union, one a completely polar and the other a com-
pletely non-polar type, and must we assume that a substance which
appears to have intermediate properties. and to be slightly lx>larized
electrically, is merely a mixture of polar and non-polar molecules?
Or can we find some means of ascribing all the most varied types of
chemical union to one and the same fundamental cause, differing only
in the nature anti degree of it~ manifestation? These are questions
which will occupy ottr attention in later chapters.
Chapter II.
The Periodic Law and the Chemist's
Picture of the Atom.
Even before the advent of the atomic theory it was known that
the elements form natural groups or families, and as soon as Dalton's
theory was established it became a matter of interest to see what
relations might exist hetween the properties of similar atoms and
their atomic weights. Ahout the same time that Prout announced the
hypothesis which gave such an impetus to the accurate determination of
atomic weights it was discovered by D6bereiner (1816) that in a
number of 'Itriads" of related elements the atomic weight of a certain
element of the triad was approximately the mean of the atomic weights
of the other two. \Ve now know many of the elements, which he
studied, to be mixtures of isotopes, and yet these coincidences pointed
out by Dobereiner are still evident in our present table of atomic
weights and have not as yet been explained.
The periodic relations bet ween the atomic weights and the proper-
ties of the elements could hardly be discovered during the period in
which many elements were assigned atomic weights which were multi-
ples or sub-multiples of their true values, but after the introduction of
the modern system of atomic weights by Cannizzaro (1858) many
chemists began to discern snch periodic relations. Probably the first
to publish anything like our present periodic table was de Chancourtois
(1863), who arranged the elements in a spiral in the order 01 their
atomic weights, and made the significant remark, "The properties of
substances aTC the properties of numbers." 5imjJar observations were
made by Newlands (1863) and more fully by Lothar Meyer (I870).
but it is to Mendel<§eff (I86<) that we owe the fullest recognition of
the periodic law and its consequences. It is unnecessary to recite here
the achievements of the periodic law of Mendeleeff. which for fifty
years has been the guiding principle of systematic chemistry. The con-
fidence in this principle was not shaken but rather strengthened by the
discovery of a completely new family of the elements, the gases of the
argon type.
Howeverf we must call attention to a ·certain error in the original
statement of the principle. The idea that the properties of the ele-
ments vary in a regular manner with the atomic weights is untenable,
for in spite of many efforts no quantitative relations have been found
between tbe atomic weight of an element and its chemical properties.
23
24 V AL),.,CE AND THE STRUCTURE OF ATOMS AND MOLECULES
If we glance at the table of the elements we see that there is a differenoe

of 3.4 units in atomic weight between sulfur and chlorine, while there is
a difference of only 0.7 between selenium and bromine. It is therefore
not altogether surprising to find that iodine has an atomic weight actually
below that of tellurium. If the elements were to be arranged strictly
In the order of increasing atomic weights, iodine would be forced into
the group of oxygen and sulfur, while tellurium would fall among the
FrequenCY ---t>
FIG. I.-X-Ray Emission Lines.
halogens. So also the positions of argon and potassium and of cobalt

and nickel would be reversed.
It was Rydberg (1897, 1914) who first comprehended the under-
lying truth in the periodic classification. The properties of an element
are determined by a single "independent variable" which is not, how-
ever, the atomic weight. In the second of his remarkable papers in
which Rydberg gave the ordinal number of each element, he was
obliged to decide upon the exact placing of all the elements of the rare
earths, upon the number of elements still remaining undiscovered, and
upon the exact position of these vacancies in the periodic table. In
all of these difficult tasks he was completely snccessfnl, and his table
of ordinal numbers is identical with our present table of atomic num-
bers, except that he assumed the existence of two elements between
hydrogen and helium. I~l of his numbers, except the'
~~
PERIOIJIC LAW AND CHEMIST'S PICTURE OF THE ATOM 25
ATOMIC NUMBERS.
~~d~1i~~m'::: :':.~:: ~
Hydrogen ....•.... Germanium ...•.... 32
Helium ............ 2 Arsenic. . . .. . . . . ... 33
Lithium ...... ... .. . 3 Selenium. . . ... . . ... 34 Terbium ....•.•.... 65
Beryllium .......... 4 Bromine. .. .. . . . .. . 35 Dysprosium. .. . . . .. 66
Boron ........•.... 5 Krypton. . . . . . . . . . . 36 Holmium .......... 67
Carbon ...•........ 6 Rubidium. . . . .. . ... 37 Erbium ........... 68
Thulium ........... 69
~~;~~~n.::::. :::::: ~ ~~~ri:!:~~.:::::::::: ~~ Ytterbium .......... 70
F1uorine ........... 9 Zirconium ..... ' . .. 40 Lutecium ........... 71
Neon .............. 10 Columbium........ 41 7·
Sodium ....•....... II Molybdenum. . . . . .. 42 Ta~t~i~m :::::::::: 73
Tungsten .......... .
~~~~~':nm ::::::::: ~~ Ruth~~{um..... .... ~
74
75
Silicon 14 Rhodium ....... "" 45 Osmium ::::::::::: 76
Phosphorus ........ 15 Palladium. .. .... ... 46 Iridium ........... . 77
Sulfur. .. .. .. . . . . . 16 Silver .......... ' . .. 47 Platinum ." , ...... . 78
Chlorine ........... 17 Cadmium ....... , .. 48 Gold .............. . 79
Argon ............. 18 Indium. . . . . .. . . ... 49 Mercury .......... . 80
Potassium .......... 19 Tin ............. , . . 50 Thallium ...... , ... . S,
Calcium 20 Antimony. . . . . . . . .. 51 Lead ............•.. 8.
Scandium .......... 21 Tellurium. . . .... ... 52 Bismuth .......... . 83
Titanium ........... 22 Iodine. . . . . . . . . . . .. 53 Polonium ......... . 84
~
Vanadium ......... 23 Xenon. . . . . ... . . . .. 54
Chromium ......... 24 Cesium. . . . . . . . . .. . 55 Nit~~·· ... ::.::::::::
Manganese ......... .25 Barium S6 87
Iron ...........•... :z6 Lanthanum. . . . . . .. 57 Radi~~ :::::::::::: 88
Cobalt ............. Z7 Cerium S8 59
~!~~~ .:::::::::::: ~ ~~=~u~.:.:::: ~
Thorium 90
9'
Z~ ............... ~ fu Uranium ::::::::::: 92
Gallium ........... 31 Samarium 62
first, by two we obtain the accompanying table, which gives the atomic
numbers that are now adopted.
The problem was also being attacked by the physicists. Rutherford
(19II) had been led by experiments on the rebounding of alpha particles
from other atoms to conclude that there lies in the center of each atom
a small nucleus, with a positive charge which can be neutralized by
the presence of an integral number of negative electrons. It was
proposed by van den Brock (1913) that the integral number which
represents the positive charge on the nucleus or
an atom represents also
Atomic numh('r--+
FIG. 2.-Atomic KU011){'r and X-Ray Frcqu('ocy.
the ordinal numher which determines the position of the element in the
periodic table.
This idea was brought into sharp relief by the extraordinary results
of the -experiments of Moseley (1913. 1914) on the X-ray spectra of
the several elements. Moseley found that whert the various elements
are bombarded by electrons in an X -ray tube each element emits a
characteristic spectrum composed of a number of high frequency lines.
These Jines are arranged in groups which appear to be identical in fann
in neighboring elements, except that they aTe shifted step by step with
the atomic number. Thus Figure I shows the wave lengths of a pair
of lines of the highest frequency, the K~ and Kp lines, of the elements
arsenic, selenium, bromine, rubidium and strontium. The evident gap
between bromine and rubidium shows a missing element, which in this
case is not unknown, but is the element krypton which cannot he made
the target in ap,.X-ray tube.
The way in which the position of anyone line cbanges from
element to element is shown in Figure 2, in which ~he atomic numbers
of the elements are plotted against the square root of the frequency
PERIODIC LAW AND CHEMISTS PICTURE OF THE ATOM 27
of a given line, the K II. line; the points faUing, within the very small
limits of experimental error, upon a continuous curve which is nearly
a straight line.
In obtaining his ordinal numbers, Rydberg conduded that there
were 32 elements in the group beginning with cesium and ending with
niton. It had previously been supposed that a larger number I probahly
36, would be found in this period. This new conclusion, which is
entirely verified by the work of Moseley, led Rydberg to a simple
arithmetical rule, which he called the rule of "quadratic groups." Since
he assumed the existence of two unknown elements between hydrogen
and helium, he arranged the elements in the following periods:
H-(?), 2; (?)-He, 2; Li-Ne, 8; Na-A, 8; K-Kr, 18;
Rb - X, 18; Cs - Nt, 32; (?) - ( ?), 32. This gives two periods of
2, two of 8, hvo of 18, and two of 32, and the numbers 2, 8, 18, 32
are equal to 2 X 1 2 ,2 X 2 2 , 2 X 32 and 2 X 42 ,
Now Rydberg was unquestionably '\... rong in assuming the two
atomic numbers between those of hydrogen and helium. This seems
to be entirely demonstrated by the relations between the spectrum of
hydrogen and the enhanced spectrum of hetiull"l., which we shall dis-
cuss in the .next chapter. Moreoyer, although only a few members
of the last period of the elements are knovm, the first part of this
period does not seem to be analogous to the period of 32 just preced~
ing, but rather shows great resemblance to the period before that, which
is one of 18. Thus thorium is more like zirconium than like cerium,
while uranium, the sixth member of the last period, belongs definitely
in the same family as molybdenum (which is the sixth member of the
last period of 18) and seems to bear no resemblance to neodymium
(which is the sixth member of the period of 32). While therefore the
facts do not substantiate Rydberg's thoory in full, nevertheless we shall
see later that his series of quadratic numbers plays an important role
in our present theory of atomic structure.
We may summarize the essential features of the periodic classifica-
tion as follows: (a) The properties of the elements are periodic func-
tions of the atomic numbers. (b) When the elements are arranged
by atomic number they fall into one period of 2 elements, two periods
of 8, two of 18, one of 32, and a fragmentary period which as far as
it is known seems to resemble a period of 18. (c) Elements which
occupy corresponding positions in the several periods have similar
properties.
Countless attempts have been made to express the periodic relation-
ships of the elements in the form of a table, a diagram, or a space
modeL Of these none can be regarded as thoroughly satisfactory.
Some fail to show all the interesting relationships which exist, others
suggest non-existing or merely formal relationships. On the whole it
seems best to employ a simple table which tells less than the whole
truth rather than more. Such a table, for which I am largely indebted
to Professor Bray, is given below. It brings out the essential relations
between the elements, although not all the interesting ones.
:z8 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
1l ~~
.S ~
§~
~-"
UiJl~N~~
d:
.
",:;;:,x><j-l<
~~l1J:i~~
;:s£::.:iilul
~z<~~~
:Z:~u~-I
3
~ OVJr.X~-tc -"
><
E
Ii
.
...l
'"<:
f-<
u
Z~<~~
0"-"
",UJo..
. ~
f-<
~ .~
'"il
0
80
~
0
O! J~~ ~
w ... .~
0..
t!Ji3~ P
ii
~ ,,=
u<:<:
-"
f-< 0
-5
'" '0
'-"0.-
zo..o.. 'w"
~
C
.2
.~
8 g,
8P1~ Vl
I Jl
,,=~ ~
1><P40 Z g
'"
~
~
::ill I
1l~:S:;, ~g
0
"
.9
1
"
.c
t.
'.g
>e~-I< t3
PERIOlJIC LAW AND CHEMIST'S PICTURE OF THE ATOM 29
Thus the relation of magnesium to zinc is not clearly indicated~ and

hydrogen might equally well be placed above lithium as above the
halogens, although this is a matter to which we shall recur.
Some Atomic Models.

In the year 1902 (while I was attempting to explain to an ele-
mentary class in chemistry some of the ideas involved in the periodic
law) becoming interested in the new theory of the electron, and com-
bining this idea with those which are implied in the periodic classifica-
tion, I formed an idea of the inner structure of the atom which,
FIG. 3.-Lewis: Memorandum of 1902.
although it contained certain crudities, I have ever since regarded as

representing essentially the arrangement of electrons in the atom. In
Fi,gure 3 is reproduced a portion of my memorandum of March 28,
1902, which illustrates the theory.
The main features of this theory of atomic strncture are as follows:
(I) The electrons in an atom are arranged in concentric cubes.
(2) A neutral atom of each element contains one more electron

than a neutral atom of the element next preceding.
(3) The cube of 8 electrons is reached in the atoms of the rare
gases, and this cuhe becomes in some sense the kernel about WhICh the
larger cllLe of electrons of the next period is built.
(4) The electrons of an outer incomplete cube may be given to
another atom, as in Mg++, or enough electrons may be taken from other
atoms to complete the cube, as in Cl-, thus accounting for "positive and
ne;:.,rative yalel1ce."
In accordance with the idea of Mendeleeff, that hydrogen is the
first member of a full period, I erroneously assumed helium to ha,'c a
shell of eight electrons. Regarding the disposition of the positive
charge which balanced the electrons in the neutral atom, my ideas \vere
very vague; I believe I inclined at that time toward the idea that the
positive charge was also made trp of discrete particles, the locali7.ation
of which dctcrmined the localization of the electrons.
These hypotheses regarding the arrangement of electrons in the
atom, while they were discussed freely with my colleagues and in my
classes, were given no further publicity. Indeed while this theory of
structure ~eemed to offer a remarkably simple and satisfactory explana-
tion of the process which occurs when sodium combines with chlorine
to form sodium chloride, it did not seem to explain chemical combina-
tions of a less polar type, such as occur in the hydrocarbons.
Yet I could not bring myself to believe in two distinct kinds of
chemical union. It seemed rather that the union of sodium and chlorine
and the union of hydrogen and carbon must represent extreme types of
a method of combination whicll ultimately would be found to be com-
mon to aU kinds of compound~. However, it was many years before
I found it possible to reconcile this idea entirely with the idea of the
cubical atom.
The first publication which recognized the stability of the group
of eight electrons was by Abegg (I904), whose paper on "Valence
and the Periodic System; Attempt at a Theory of Mo1ecular Com-
pounds" ends with the significant remark, "The sum 8 of our normal
and contra-valences possesses therefore simple significance as the num-
ber which for all atoms represents the points of attack of electrons;
and the group-number or positive valence indicates how many of the
8 points of attack must hold electrons in order to make the element
e1ectricaUy neutra1."
The next important contribution to the interpretation of the periodic
law was made by J. J. Thomson (I904) who considered the mathe-
matical consequences of the assumption that the atoms of the elements
consist of a number of electrons 4lenc1osed in a sphere of uniform
positive electrification." He was thus led to conclude that a ring of
electrons, eqw.aUy spaced and revolving about a positive center J would
be stable nntil the number of electrons in the ring exceeded a certain
number, and would thell break into two concentric rings. Thus, if the
nnmber of electrons in the outer ring is increased, a limit is reached
PERIOIYlC LAW AND CHEMIST'S PICTURE OF THE ATOM 31
where another ring is formed, and so on. As an illustration of the

various types of stability, he showed that if a numher of small magnets
are floated by corks upon a surface of water, so that all of the north
poles point upward, and if the south pole of a larger magnet is brought
near the surface. the sll:j:aU magnets will orient themselves ahout the
larger one in concentric rings. Thomson recognized that under certain
circumstances electrons would arrange themselves not in rings in a
certain plane but in pclyhedral figures about the center, but the difficulty
of mathematical calculation in such cases led him to restrict his atten-
tion to the arrangements in a single plane, and this decision may have
been partly responsible for some later theories which assume a co-planar
arrangement of electrons in atoms.
Thomson saw immediately the analogy between his arrangement of
electrons and the periodic system of IVfendeleeff. "Thus if we consider
the series of arrangements of corpuAcles (electrons) having on the
outside a ring containing a constant number of corpuscles, we have, at
the beginning and end, systems which behave like the atoms of an
element whose atoms are incapahle of retaining a charge of either
positive or negative electricity; then (proceeding in the order of increas-
ing number of corpuscles) we have first a system which behaves like
the atom of a monovalent electropositive element, next one which
behaves like the atom of a divalent electropositive element, while at
the other end of the series we have a system which behaves like an atom
with no. valency, immediately preceding this, one which behaves like the
atom of a monovalent electronegative element, while this again is
preceded by one behaving like the atom of a divalent electronegative
element.
"This sequence of properties is very like that observed in the case
of the atoms of the elements. Thus we have the series of elements:
He Li Be B C N 0 F Ne
Nt Na l\fg AI Si P S CI A
"The first and last element in each of these series has no valency,
the second is a monovalent electropositive element, the last but one
is a monovalent electronegative element, the third is a divalent electro-
positive element, the last but two a divalent electronegative element,
and so on.
"When atoms like the electronegative ones, in which the corpuscles
are very stable, are mixed with atoms like the electropositive ones, in
which the corpuscles are not nearly so firmly held, the forces to which
the corpuscles are subject by the action of the atoms upon each other
may result in the detachment of corpuscles from the electropositive
atoms and their tra.nsference to the electronegative. The electro-
negative atoms will thus get a charge of negative electricity, the elec-
tropositive atoms one of positive, the oppositely charged atoms will
attract each other, and a chemical compound of the electropositive and
electronegative atoms will be formed." '
It is evident that Thomson's picture of the union of two atoms
is entirely similar to the one offered by Abegg, although unlike the

latter Thomson considered his results as partly deducibJe from certain
laws of force, corresponding to his assumption of a sphere of positive
electrification in which the electrons were supposed to be imbedded.
Such an idea regarding the positive part of the atom soon proved to
be untenable, Rutherford's study of the scattering of alpha rays by
various substances seems explicable only on the assumption that the
positive part of an atom is concentrated in a very small region at the
atomic center, and he therefore proposed what may be called the
planetary theory of atomk structure, accordjng to whicb the electrons
were assumed to be revolving in orbits about the small positive nucleus,
and subject to the same laws of force (the inverse square law) that
governs the motion of the planets about the sun. This theory of the
planetary atom will be further discussed in the next chapter.
In 1915 Parson published a very interesting paper entitled "A
Magneton Theory of the Structure of the Atom." Here the electron
was regarded as a rotating ring of negative electricity which therefore
possesses a magnetic moment 1 and could be called a magneton. As
in Thomson's theory, the electrons or magnetons were assumed to lie
inside a large sphere of uniform positive electrification, and Parson
beHeved that the magnetic forces between the magnetnns would cause
them to arrange themselves in cubes (not concentric, but lying along-
side one another in the large positive sphere).
There is one feature of Parson·~ theory which is now generally
regarded as erroneous, and for which I am afraid that I am in part
responsible. When Mr. Parson first showed me his magneton theory
he considered his rings of electricity capable of various degrees of
velocity, sometimes even exceeding the velocity of light. It was at
my suggestion that he attributed a nxeri magnetic moment to his
magneton so as to make it in a sense the elementary unit of magnetism
as the electron is also the unit of electric charge. This idea has not
proved fruitful, and it seems unlikely, although perhaps not impossible,
that an electron possesses any magnetic properties except when it
is a part of an atom or a molecule. It is, however, to be observed
that in the Bohr theory, which we shall discuss in the next chapter, and
which offers a more satisfactory picture of the motion of electrons
within the atom than Parson's theory, there appears again a definite
unit of magnetic moment.
Parson's paper was largely devoted to a discussion of the stability
of the groups of eight electrons and the tendency to form such groups
in various types of chemical union, and he showed that those com-
pounds in which these complet:e groups of eight cannot be assumed
are the ones whose magnetic properties show that the molecule possesses
a large magnetic moment.
In March and in April 1916 there appeared two papers, one by
:I A brief description of some of the elementary principles of magnetism will

be given in the next cbapter.
PERIODlC LAW AND CHEMlST'S PECTURE OF THE ATOM 33
Kossel, "'Cber Molekiilbildung aIs Frage des Atomhaus," and onc by

myself on HThe Atom and the :Molecule." These two papers offered
closely parallel pictures of the structure ot the atnms and ot those
molecules of the more polar sort, where each atom may be regarded
as existing in the state of an ion. In both papers the electrons of
an atom were regarded as surrounding the sman positiyc nucleus in
concentric groups, the fin,t being a group of two, the second a group
of eight, the third a group of eight, and then other groups of some-
what indeterminate character, but always ending in ail outer group
of eight electrons in the atoms of the rare gases, as also in simple
elementary ions.
Kossel assumed these successive groups to OCCllr in concentric rings
about the nucleus, while I (in accordance with my early views ex-
J J
I)
EJ II
Lewis, 1916 Kosse1.19 16
FIG. 4.~ Two Models of the Argon Atom.
pressed ill Figure I) assumed these groups to constitute concentric

shells forming a three-dimensional structure about the central atom.
l. ~igure
. 4 shows the t\'vo pictures of the argon atom side by side.
In both theories the electron groups were supposed to reach the
highest degree of symmetry and stability in the atoms of the se:'1eral
rare gases; helium with its group of two, neon with its groups of two
and eight, argon with its groups of two, eight and eight, and so on.
Other atoms were concel\"ed as having a strong tendency either to give
up electrons or to take up electrons in such manner as to ape the
structure of the nearest rare gas.
In this connection I emphasized the peculiarity of hydrogen which,
by giving off an electron, can become the simplest of positive ions,
~onsisting solely of an atomic nucleus, while by taking on one electron
1t can complete the group of two, characteristic of the helium atom.
This process seemed so nearly like the taking on of one electron by
flnorine or by chlorine to form F- or Cl-, with structures corresponding
to neon and argon, that I felt justified in regarding hydrogen as
~longing, in this respect at least, to the halogens; and therefore pre-
dicted that the metallic hydrides would prove to have the character
34 VALENCE AND THE STRUC{URE OF ATOMS AND MOLECULES
of salts, consisting of metal ion and Ilydride ion, and further that
electrolysis of a hydride should produce hydrogen at the anode. This
prediction has been entirely verified in the work of Bardwell (1922)
who succeeded in electrolysin~ a melt containing calcium hydride, and
obtaining hydrogen at the anode in amount corresponding to Faraday's
law.
Aside from this one case, atoms show a vcry marked tendency to
form an outer group of eight electrons, and this tendency furnishes a
very simple interpretation of a large class of the morc l)oIar chemical
compounds, as shown in the scycral pal)Crs of Parson, Kossel and
myself. :My paper went further and attempted to furnish an equally
simple explanation of compounds of the less polar type, but this will
be the subject of a later chapter.
Chapter III.
Spectral Series and the Physicist's View of the Atom.
As soon as the kind of light emitted or absorbed by a substance was
rc{:ognized to be peculiar to that substance, the 5.tucly of characteristic
spectra became one of the important methotls of chemical analysis. It
was Kirchhoff and Bunsen (1860, 1861) who showed the great power
of the new method, not only in the detection of existing elements,
but also in the disco\"cry of new elernents. The heavy alkali metals,
rubidium and cesium, were thus discovered by them.
Indeed when metallic salts are placed in a flame the bright spectral
lines which are obtained seem ordinarily to be characteristic of the
metal rather than of the particular compound which is used. It has
long been recognized that the various emission spectra, whether fr0111
arc, or spark, or Geissler tuhe, or Hame, fall into two classes, which
are technically known as line spectra and band spectra, and there has
been no occasion to abandon the view, first suggested hy Helmholtz,
that while the banu spectra are characteristic of molecules, the line
spectra are due to atoms which have been set free under the condi-
tions which give rise to the emission of light.
Now it is evident that the characteristic spectral lines of the ele-
ments, which afe so readily studied, and whose wave lengths can be
determined with an accuracy which is hardly attained in any other type
of physico-chemical measurement, should furnish information of great
value concerning the inner structure and behavior of the atom. But
before this information can be utilized we must have some theory of
the way in which light is emitted or absorbed. by a substance.
The undulatory theory of light strengthened the analogy between
light and sound. lvlonochromatic light is characterized by its fre-
quency, or wave length, just as a musical tone is characterized by
its frequency, or wave length in air. The emission of a musical tone
is d~e to something which is vibrating, like a tuning fork. So, ac-
~ordmg to what we may can the classical theory of light emission, light
IS due to the vibration of something within the molecule or atom, and
after the adoption of Maxwell's electromagnetic theory, it was assumed
that the something vibrating carried an electric charge.
According to this classical theory the elementary vibrators or res-
onators obey the familiar laws of elastic bodies and thus possess a
r:atur,al frequency. or pe~od, independent of the amplitude of vibra~
tlOn If such amplItude IS small. These vibrators, set in motion by
35
thermal action Of by electric discharge, are then the source of light;

and the emission of a bright spectral line is due to the preponderance
of vibrators possessing some one natural frequency.
Conversely when 11ght falls upon a substance containing such
vibrators the latter are supposed to acquire energy of vibration at the
expense of the light, and especially of that part of the light which has
a frequency corresponding to the natural frequency of the vibrators.
The energy of vibration thus acquired is then converted into thermal
energy. This is the classical theory of light absorption.
Now this theory of the emission and absorption of light through
the vibration of charged parts of the molecule or atom has furnished
a very satisfactory explanation of a large number of phenomena. It
is therefore with some regret that we now find ourselves obliged to
give up in large part, if not wholly, this simple picture of the inter-
action between matter and light.
Some evidence of the inadequacy of the classical theory is furnished
by the occurrence of spectral lines 111 groups or ser1cs. Now by analogy
to a musical instrument that sends out a series of tones and over-
tones it was to be expected that the elementary vibrators, especially
if they exert a mutual influence one upon another, might emit not
one but a series of bright lines, anJ the discovery that a single element
does emit a whole series of spectral lines see~ed at first to support
such an analogy. But the quantitative relation between the frequencies
of the several lines of an elementary spectrum proved to be very differ-
ent from anything that was to be expected from the analogy to musical
tones.
Various attempts to express in a simple numerical formula the
several lines in a single spectral series "",ere unsuccessful until Balmer
( 1885) obtained for the important hydrogen series a formula which
thenceforth became the prototype of all formu1re for series of line
spectra. This formula of Balmer, although containing but a single
arbitrary constant, reproduced with marvellous accuracy the positions
of the lines of the hydrogen series as they had been observed, not
only in the laboratory, but also in the spectrum of sun and stars.
Balmer expressed the series of lines by the formula
(I)
where A is the wave length and n is anyone of the whole series of

integers from 3 to 8. Each integral value thus corresponds to a single
line of an infinite series in which the lines becotne closer as n increases,
and converge at a limiting value known as the head of the series, where
~=A. .
The frequency of a given line (in reciprocal seconds) is equal to
clA, and therefore the equation may be written
(2)
SPECTRAL SERIES AND PHYSICISTS VIEW OF THE ATOM 37
or in alternative fonn,
(3)
Modern refinements of the methods of spectroscopy have greatly in-

creased the accuracy of the determination of spectral lines. Twenty
lines of the Balmer series have been obtained in the laboratory and
thirty from stellar observation, and the frequencies of these lines do
not on the average differ from those calculated from the formula by
more than one part in a million.
One of the most significant discoveries relating to spectral series
was made by Pickering (r897) in observing the spectrum of the star
~-Pupis. Here a series was found, every alternate line of which ap-
peared to coincide with a line of the Balmer series. The whole series
was accurately represented by the formula
II'
1.=A n'_r6" (4)
This formula gives lines identical with those of the Balmer formula
when n == 6, 8, etc., and gives the additional lines of the Pickering
series when n == 5, 7, etc. These new lines were originally ascribed
to hydrogen in some peculiar form, but we shall see that Bohr has
shown this series to be due to helium, and the similarity between the
Balmer and Pickering series thus furnishes a remarkable illustration
of the intimate inter-relationship bet\veen the line spectra of different
elements.
The Work of Rydberg.
Until ten years ago the great accumulation of exact spectroscopic
data regarding line series had led to the discovery of only two' funda-
mental generalizations, both of which were made by Rydberg (lSgo),
whose wonderful perspica.city we have already recognized in his dis-
cmrery of the significance of the atomic numbers.
His first discovery was that a certain number appeared in the
arithmetical expression for the line spectra of a number of elements.
This number, No, which he announced as "a constant common to aU
series and to all elements," is now recognized as a universal constant
of great significance. This constant of Rydberg (except for a very
small correction) is the coefficient which appears in the second member
of Equation 3.
Another equally important generalization of Rydberg has become
known as the combination principle. When an element exhibits two
or more different line series, the lines of one series and those of
another are simply related to one another. According to the combina-
tion principle, as it is now usually stated, the frequency of each of
the ma!IY spectral. lines obtained !rom the same atomic species may
be obtamed by taking the several dIfferences between a relatively small
38 VALENCE AND TIlE STRUCTURE OF ATOMS AND MOLECULES
OOmber of basic frequencies. In order to il1ustrate this point we may

consider the complete spectrum of monatomic hydrogen.
In addition to the Balmer series there are a number of other im-
portant spectral lines which are ascribed to free hydrogen atoms. One
important series has been obtained by Lyman ('904, 1<)06) in the
ultra-violet, another by Pa~('hen (1909) in the ultra-red, and very
recently a few lines of a fourth series have been obtained by Brackett
(1922) in the extreme ultra-red. The formulx which reproduce the
several series are as follows:
(Lyman) v=No(~-~),
(Balmer) v= No(~-~~),
(Paschen) v=No(?-~).
(Brackett) v=No(~-~).
The first three of these series are represented in Figure 5. where only
a few of the lines of each series arc shown (together \'lith the head
of the series, represented by a dotted line).
]11 !I I I
Paschen Balmer
III
Lyman
j
FIG, 5.-Three Hydrogen Series.
Now it is evident that a 1ine of anyone of these series may be

expressed by the formula
v=N,(~-;,), (5)
therefore the frequency of every line can be regarded as the difference

between some one pair of the basic frequencies
(6)
Thus the second line of the Balmer series is v~* - v.._*, the third line
of the Lyman series is Vl* -V4*' and the first line of the Paschen
series is vs*-v.*.
It is customary to express the combination principle graphically
as in Figure 6, where again frequencies are plotted from left to right.
The vertical ·lines represent the values of v*, the basic frequencies,
and the length of the horizontal lines terminating at these vertical
lines represent the frequencies of three spectral lines that we have
just mentioned.
SPECTRAL SERIES AND PHYSICIST'S VIEW OF THE ATOM 39
\Vhen we 'turn to other elements than hydrogen we find a condi~

tion of far greater complexity. As a rule it is no longer possible to
express either the frequencies of the series lines, or the basic frequen-
cies, by any such simple fonnl1lre as sufficed in the case of hydrogen.
The equations which have oeen most useful in expressing these series
require several arbitrary constants, and appear to be of only approxi·
mate validity. Nevertheless we tind also in these more complex cases
that the observed frequencies arc Ollce more ohtainaille as exact
ll~
NoNo No
·1
No
No
°5'"77 "7
FIG. 6.-Basic Frequcilcics of Hydrogen.
differences between certain basic frequencies, in complete accordance

with the combination principle.
These ideas, which originated \vith Rydherg, have hecome the
foundation of the theory of spectral series ,,,,hich has been so rapidly
developed during the past decade. Before disC'ussing these rcccnt
advances, it win he necessary to gi \'e brief consideration to that revolu-
tion in :-,cientific thong-ht which is known as the quantum theory.
The Quantum Theory.

It was one of the foremost deductions of the kinetic theory of gases
that the molecules of all gases should possess the same average kinetic
energy of translation at a given temperature. This idea was carried
over by analogy to liquids and solids, and it was assumed that every
particle would have the same average kinetic energy at a given tempera-
ture. This is known as the Law of the Equipartition of Enerf:,'Y.
This principle led Boltzmann to the explanation of the Jaw of
Dulong and Petit. If the atoms of a solid posgess. the same kinetic
energy as the atoms of a monatomic gas, anti if they vibrate about
fixed positions in accordance with Hooke's law, so that (as in the
case of any simple vibration) the average potential energy equals the
average kinetic energy, then the total thermal energy of the atoms
in a solid would be twice as great as that of an equal number of
atoms of a monatomic gas; and the energy would increase twice as
rapidly with the temperature for the former as for the latter. But
Dulong and Petit's law is only true as a limiting law at high tempera-
tUres. Many substances at ordinary temperatures and all silbstances
at low temperatures show a very much smaller change of energy
Witl.l the temperature than that law requires. If we plot the energy
agatnst the a~olute temperature, as in Figure 7, the dotted line ex-
presses Dulong and Petit's law, \\'hiJe the behavior of an actual sub-
stance is indicated by the continuous curve, which has the same slope
as the dotted line only at high temperatures.
The theory of equipartition appears to give a very satisfactory
explanation of the thermal e-nergy of a monatomic gas. But the atom
of such a gas is by no means the ultimate particle. By the equiparti-
tion law ea('h of the several particles which compose such an ato111
should acquire its quota of energy. On the contrary we are con-
vinced that a monatomic gas acquires no appreciable thermal energy
3 00
Temperature. ahsohlh' ----I'
FIG. 7.-Hcat Content of Copper,
except that which is due to the translational motion of its molecule

as a whole.
In spite of the ohyious falsity of the equipartition law, it neverthe-
less seems to be a direct consequence of accepted mechanical principles.
Perhaps its deduction from these principles by the methods of statistical
mechanics has never been made in a manner entirely free from ob-
jection, but it is generally believed that such a deduction is possible.
Thus we are led to the inference that the mechanics of atoms differ
in some respect from the known mechanics of massive bodies.
The equipartition law fails not only in its application to the thermal
energy of ordinary bodies, but also when it is applied to the distribn-
tion of energy in the spectrum of radiation emitted by a black body.
In such radiation we presumably have light of all frequencies from
zero to infinity, and we may speak of the amount of radiant energy
which is comprised between two· chosen frequencies Vi and V2. It
was sho\vn by Rayleigh (1900) to be a consequence of the equipartition
law that the e.nergy comprised between two such fixed limits of fre-
quency must be proportional to the absolute temperature, as shown
by the dotted line in Figure 8. The continuous curve of that figure
represents the actual facts as brought out by Wien (1896) and more
SPECTRAL SERIES AND PHYSICIST'S VIEW OF TIlE ATOM 4'
fully by Planck (H)OI). This curve approaches a straight line, as

demanded by the equipartition law, only at high temperatures.
Planck, seeing that his equation for the distribution of radiant en-
ergy was incompatible with the deductions from accepted mechanics
and electromagnetic-s, announced a hypothesis of unusual holdlll'SS
which, together with the mass of new Jaws and hypotheses which have
grown out of it, is known as {he quantum theory. Planck assmned
in the first instance that bodies contain electrical oscillators which
can absorb or emit radiation, not in a continuous manner, but by finite
,
,,
,,
,l
,,
;
,
,,
I
/
,
/
,,
I
,,
,
/
,,
Temverature _
FIG. 8,-Density of Radiant Energy (Between Two Given Frequencies).
amounts; and that each of these finite amounts of energy is propor-

tional to the natural frequency of the oscillator. He further assumed
the proportionality factor to be the same for all oscillators and there-
fore a universal constant, which is denoted by It and known as the
Planck constant. According to this theory, an oscillator whose natural
frequency is v can possess energy only in the amount hv or some
multiple of hv.
Einstein (19"7) suggested that we go further, and regard the
energy, hv, emitted by an oscillator, as a quantum or corpuscle of
radiant energy which. preserving a certain degree of individuality,
could only be reabsorbed as a whole. This idea bas not been widely
accepted because of the difficulty of. its reconciliation with the phe-
42 VALENCE AND Till! STRUCTL'IIE OF ATOMS AND MOLECULES
nomenon of the interference of light. However, it led Einstein to

one of the most important deductions from quantum theory, namely,
his photoelectric equation. According to this equation, when a sub-
stance is subjected to monochromatic light of frequency v, the maxi-
mum energy which can be acquired by onc of hs electrons 15 hv.
It was also Einstein who saw the connection between the two types
of deviatiun from the equipartition law, of \vhich we have spoken. If
a solid body is supposed to be composed of atom~ which are analogouS
to the hypothetical oscillators of Planck, and therefore capable of
a.cquiring energy only in finite increments, the energy \vauld not be
the linear function of the temperature required by the law of Dulong
and Petit. At low temperatures many atoms would be unable to acquire
the energy llv, and therefore would possess no energy at all. Thus,
from the Planck radiation formula, Einstein obtained his equation for
the specific heats of solids, which qualitatively, although not quantita-
tively, is in accord \vith the numerous measurements of specific heats
that have since been made at low temperatures.
Once in the early days of quantum theory Professor Einstein re-
marked to me that the quantum theory was not really a new theory,
but merely a recognition of the falsity of previous theories. This
remark remains trUe. Some scientists have been inclined to give up
such fundamental ideas as the laws of conservation of momentum
and of eneq"ry, and to replace them hy analogous theorems which are
only yalid in a statistical sense. Others haye gone so far as to con-
dude that the continuum of space and of time must he replaced by a
discontinuu111.
For the moment l\'e need only conclude that in giving up the con-
tinuous theory of matter, and replacing it by the theory of di~crete
('enters which we call atoms (or electrons and nuclei), We have some-
how failed in consistency. A race with more limited sense perceptions
than onr own_might study the properties of sand and conclude these
properties to be due to the existence of grains, but \vollid they then
be justified in regarding the grains as composed of sand? Y~t this
is the kind of inference that modern science has sanctioned. The prop-
erties of electricity have been explained by assuming it to he com-
posed of electrons, after which we naively consider the electrons as
made up of electricity, and speculate concernina the distribution of
electricity about the electron center. \Ve also hav~ regarded the atoms
as possessing properties similar to those of the larger bodies which
t!lCY compose. The various phenomena which are grouped under the
tItle of quantum theory are the new data in the light of which we
must construct the .new geometry and the new mechanics which are
valid in the i1l1l11e,d.ia.te vicinity o~ ~]ectrons and nucJei. Quantum
theory has been crlttclzed for furmshmg no adequate mechanism but
pres,:,~ahly the' root of our present problem lies deeper than' this,
and lt IS haf(~Jy lIkely that any rnechan~srn based on our existing modes
of thought wdl suffice for the explanatIOn of the many new phenomena
which the study of the ato111 is disclosing.
SPECTRAL SERIES AND 1'11l'SlClST'S "1H/V OF 1'111£ ATmI 43
A Partial Statement of Bohr's Theory,

.O.ut of the. chaos of spectroscopy with its scarcity of guiding-
prmclples and Its abundance of almost unrorrelated data. ordtr and
simplicity were achieved at a single stroke through the hrilliant theory
proposed by ,Bohr (19I3). To this theory, which has so justly cap-
tivated the minds of physicists, we must 110\\" give oUr attention. 1t
FIG. 9.-Energy Levels in the Hydrogen Atoln.
will be presented in a somewhat different manner from that employed

originally by Bohr himself, in order to separate that part of the theory
which may be stated independently of any atomic model from the part
which deals with his model.
Considering the simplest case of an atom like monatomic hydrogen,
which is composed solely of a nucleus and one electron, it is fjrst
postulated that the electron may assume anyone of a whole series
of states, so that each state may be spoken of as an energy level. Thus
in Figure 9 we may represent these levels by a series of lines, such

that the distance betw~'en two lines is equal to the energy difference
in the atom between the two corresponding states. The lowest level
represents the smallest energy and therefore the most stable state
of the atom.
These levels form an infinite series, and the energy at the limit
of this series (indicated by the dotted line) is assumed to represent
the energy of the system when the nucleus and the electron are COI11-
pletely separated from one another. The difference between the
energy of the nth level and this limiting- energy can be designated by
En* (a negative quantity). For convenience, the value of the limiting
energy may be taken as zero, and we may then say that the energy
of the first level is E1*' that of the second level E2*' and so on. All
of these values of E* are now negative.
The second assumption made by Bohr is that the energy at any
level, when divided by h, the Planck constant, gives the negative of
one of those fundamental frequencies which we have discussed in a
preceding section, and from which the various spectral lines of the
element may be obtained. In other words, it is assumed that
(7)
The third a~sumption is that the atom emits light only when the
electron falls from one level to another (or absorbs light only when
the electron is raised from one level to another). The energy of the
emitted light is equal to the difference between the energies of the two
levels, and the frequency of the light is this energy difference divided
by h. Thus
v (8)
So the second line of the Balmer series, which is V2*-V,,*. is supposed

to be produced when the electron faUs from the fourth to the sec-
ond level of the atom of hydrogen, and the other lines of the Balmer
series are obtained when an electron drops from the several upper
levels to the second level; while the lines of the Lyman series are
produced when the electron drops from various levels to the first and
most stable level. •
'iVithout going more fully into the complex spectra of the elements
than has been possible in this brief summary, it is impossible to show
what a wonderful insight into the significance of spectral series is
furnished by Bohr's theory. Nor does this partial statement of his
theory alone do justice to that brilliant generalization, lOT when we
consider Bohr's . ,atomic model we shall see that an equal1y simple
set of assumptions leads to more far-reaching conclusions than can
be drawn from tne assumptions which we have so .far discussed.
However, it will be seen that the theory as we have developed it
so far contains the two essential elements of the quantum theory. The
SPECTRAL SERIES AND PHYSICIST'S VIEW OF TIlE ATOM 45
first of these is that monatomic light of frequency v can change the

energy of an electron by the amount hv. The second is Planck's
idea that there is a mechanism in the atom whose energy can vary, not
continuously, but only by finite increments.
The Interpretation of X-Ray Spectra.

The theory of Bohr not only explains the long familiar data of
spectroscopy, but by a slight modification gives an equally satisfac-
tory interpretation of the newer data relating to X-ray spectra. Let
us consider a heavy element whos;e atom contains a large number of
electrons, and postulate once more that there aTe certain energy levels,
the first being now called the K level, the next the L level, the next
the M level, and so on. We may next assume that only a limited
number of electrons can be present at each of these levels. Instead
of considering these levels as a mere framework, as we do in con-
sidering an atom like hydrogen, we may assume that each of the
levels-at least each of the lower levels-contains its full quota of
electrons. If then bv some means an electron from the K level is
knocked out of the atom, an electron from the L level may fall in
to take its place, thus giving rise to the spectral line desiinated as
K". If it is an elee-tron from the 11 level which falls into the vacant
place it gives rise to the line K'f,I, and electrons dropping from higher
levels give a bundle of lines which, if the spectroscopic resolution is
not great, are usually observed as a single line. K'Y' So also an
electron may be ejected from the L level. and if its place is filled
by an electron from the .11 level the L" line results, and so on, for the
other lines.
This theory furnishes a beautiful explanation of the peculiar ab-
sorption of X-rays. \Ve do not find, as in 'the case of ordinary spee-tra,
that \""e have absorption lines corresponding to the several emission
lines. The absorption of X-rays at the frequency K" would indicate
an electron being trans ferred from the K leyel to the L level, but in
the stable atom the L level already has its quota of electrons, and this
process is therefore impossible. Absorption cannot begin until the
frequency of the X-rays is sufficient to remove the electron entirely
from the atom, or at least to one of the outer levels which has not
its full quota of electrons. The observed fact is that a continuous
absorption band stretches from the higher frequencies down to a
frequency a little higher than corresponds to the K, line and there
abruptly terminates.
If the energy at a given level in different elements is supposed
to depend primarily upon the charge of the nucleus, we have a new
interpretation of the simple relation between the X-ray spectra and
the atomic number which }\rIoseley discovered, and which has been
illustrated in Figure 2 (Chapter II).
. This simple concept_ion has. b~en of the greatest service in account-
mg for the general charactenstlcs of X-rays, although here as with
46 VALENCL .1ND 1'11E STRUCTURE OF ATOMS AND MOLECULES
onlinary spectral lines there are certain complexities, a discussion of

which would carry us too far afield. The concept of energy levels
requires that the comhination principle, which proved so useful in
the understanding of ordinary spectra, should hold for X-ray spectra
also. Thus if the line Kp is determined by the energy difference
hetween the K and 1\1 levels, and the Kn line by the difference between
the K and L levels, while the Lo. line is determined by the difference
between the L amI 1\1 lc\'els, the frequency of the K" line should
be the difference Lctween the frequencies of the other two, and this
i~ an experimental fact.
Ionization and Resonance Potentials.

In a vacuum tube \\then the electrons emitted hy a hot cathode
pass through a potential gradient to an anode, the electrons acquire a
kinetic energy corresponding to the difference 1n potential between the
cathode and anode. Also when certain gases are introduced into the
tube, the electrons, although meeting and rebounding fro111 the gas
molecules, seem to do so in an elastic manner, and finally arrive at
the anode with the same kinetic energy which they would acquire in
the absence of any bras. This is far from being true of all gases, but
there are many in which these collisions between electrons and mole-
cules appear to he completely elastic within the limits of experimental
error. In other words, the motion of the electron through the gas
can be said to be frictionless.
nut even ,,,,ith gases of this type, if the potential difference between
cathode and anode is g-radually increased. a definite point is reached
at which the electron evidently loses energy upon collision. Vve may
say that the slow-modng electron rebounds from the molecule elastically,
but that when the kinetic energy of the electron reaches a certain
value, a part of its energy is given ttp to the molecule on collision. This
important ohserYation, first made by Franck and Herz (1913), has
been confirmed in many investigations.
If the kinetic energy of the electron is increased beyond this first
point of inelastic collision, other points become manifest, which indicate
new processes hy which the energy of the electron is given to some
part of the gas molecule. The first of these critical points to be
ohserved is often associated with the sudden emission of light. The
frequency of the light is that of one of the characteristic spectral lines
of the gas. The potential which suffices to produce this inelastic col-
lision, accompanied by light emission, is known as a resonance potential.
In the simpler cases, the highest .of these critical potentials is found
to be associated with the sudden appearance of gaseous ionization.
It appears therefore that the electron striking the molecule knocks
off another elechon, so that the two electrons departing leave behind
a positively charged ion. The minimum potential required to produce
this phenomenon is known as the ionization potential.
Bohr's theory gives an extremely satisfactory qualitative and quan-
SPECTRAL SERIES AND PHYSICIST'S VIEW OF THE ATOM 47
titative explanation of these phenomena. If we consider an atom

with an electron in the most stable position, that is, upon the lowest
available energy level. the atom can only acquire energy if enough
is furnished to raise the electron to one of the higher energy lc\·cls.
Thus if the atom is struck by an electron which has insufficient energy
to produce this result the collision must be elastic On the other
hand, if the bombarding electron possesses just enough energy to raise
the electron within the atom to the next energy level, it may lose
all its kinetic energy. Then the collision is inelastic, and the electron
which has been raised to the second level, or resonated, maya moment
later drop back to the first level, emitting the first line of the element's
series spectrum. Again, the velocity of the bombarding electron may
be great enough to cause the electron of the atom to pass through the
whole series of energy levels and become separated altogether from
the atom, which therefore becomes ionized. Such is the qualitative
explanation of resonance and ionization.
The quantitative explanation is equally satisfactory. The energy
required for the first resonance must be equal to h times the frequency
of the first line of the spectral series. The energy required for ioniza~
tion must be h times the limiting frequency of the series (the head of
the series). These deductions from the theory have been completely
verified by experiments with a large number of elements, within the
limits of the experimental accuracy, which unfortunately is not yet all
that might be wished.
The phenomenon of resonance and ionization potentials is most
sharply evidenced in the metallic vapors. In the case of hydrogen the
presence of diatomic molecules complicates the situation, hut if we could
study pure monatomic hydrogen we should doubtless find the first reso-
nance JXltential and the ionization potential to be in the ratio of 3 to 4,
since the former depends upon the term (~-~) = J., and the latter
I 2 4
upon the term (~-~)== I, thus corresponding to the first line, and
to the head, of the Lyman series.
These experiments on resonance and ionization potentials seem to
furnish a complete demonstration of the quantum assumption that
definite energy levels exist within the atom, an(~ that an electron can-
not be lifted above one level unless it receive a sufficient amount
of energy to raise it completely to another level.
Bohr's Atomic Model.
Having seen how exceedingly useful even a partial statement of

Bohr's theory can be made, let us tum to his complete theory of the
structure of the hydrogen atom. He assumed first a Rutherford atom
with a small positive nucleus and an electron revolving about it in
a circular orbit. The centripetal force is taken as that given by
48 VALENCE AND THE STRUCTUIIB OF ATOMS AND MOLECULES
Coulomh's Jaw, namely, a force C(lual to the product of the two

dlarges and inversely proportional to the square of the distance be-
tween them. The properties of the system afC in this respect identical
\vjth those of a ~)'stem composed of the sun, and a planet moving
in a circular orhit.
In C'onfOfmitv with this In.w of force a continuous series of orbits
is possible, the radius of each orhit determining the yclacity in the orbit,
and the energy (kinetic anti potential) of the system. Now Bohr in-
troduced the (luantulll theory by assuming that not all of these orbits
FIG. JO.-Orbits of the Hydrogen Atom (Bohr).
are possible, hut only that particular set of orbits in which the angular
momentum of the electron is an integral multiple of h/21C (Figure 10).
The" orbit nearest the nucleus is the one in which the angular mo-
mentum has this value. the second is one in which it has twi<'e
this value, and so on. In conformity with the assumed law of force
these orbits must have radii in the ratio of 1:4:9 :16, and so on
to infinity, the radius of the first orhit being of the order of 10-$ cm.
His third assumption is the one which we have discussed before,
namely. that emission of light occurs when the electron falls from
one orbit into an inner orbit, and that the frequency of the emitted
light is equal to the difference in energy between the two orbits (or
energy levels).
By means of this astonishingly simple assumption, it is possible
SPECTR.;L SERIES AND PHYSICIST'S r}l;JV OF THE .HWI 49
to obtain quantitatively the whole set of spectral serie~ which we as-

cribe to atomic hydrogen. Aferely from Coulomh's law and the
assumption that the angular momentttm in each orhit is a l~1Ultiple 0\
11/21(, we may (_'alcnlate the energy of the atom corrcspondmg to any
orbit. From this calculation,
(<»)
where {" is the charge of the nucleus. I' the charge on the electron.
m the mass of the electroll aneI h the Planck constant.
For the hydrogen atom r' =::: c, amI if we u"e Equatl011 g we fill(1
for the frequency of any line
(10)
Kat only does thi:::. equation resemble in form the (''fJl1atioll hy \vhich
we have already expressed the spectral lines of monatomic hydrogen,
but the coefficient in the last term of E<luation 10 should be equal
to the Rydherg constant, and it is so within the narrmv limits of error
involved in the determination of the several quantities concerned.
Bohr next ~h()wed that an, atom with a nuclear charge equal to
tWIce that of hydrogen, and possessing a single electron, should give
spectral lines according to the formula
(II)
This formula satisfied the Pickering series, as well as certain other

spectral lines which had formerly been ascribed to hydrogen, but which
Bohr showed must be due to helium atoms from which one of the
two electrons had been ejected, 'namely. to the ion He+. This in~
ference has since been completely verified by the production of these
lines in tubes containing pure helium.
As a matter of fact, the values of Rydberg's constant obtained from
the Balmer series and from the "enhanced" helium series are not
ahsolutely identica1. Although the difference is small, the methods of
spectroscopy have been so refined that it can be measured with some
accuracy. This difference was also readily explained, since the mass
of an electron is not entirely negligible with respect to the mass of
the nucleus of hydrogen or helium, and therefore instead of assumiqg
a system in which the nucleus occupies a fixed position and the electron
revolves about it, the two must be considered as revolving about their
common center of mass, which is very close to the center of the nucleus.
In~eed, by making use of the lmown masses of the hydrogen and
helium atoms and the two values of the constant in the equations for
the spectral series, it has been found possible to calculate the mass
of an electron with an accuracy which apparently rivals that given
by other methods.
It would carry us too far to attempt to describe the many interest-

ing developments and refinements which have been introduced into
Bohr's theory by the assumption of elliptical as well as circular orbits,
the eccentrjcity of the orbits varying not continuously but in steps,
in accordance with a second application of quantum theory. Many of
the morc minute observations regarding spectral lines to which we
have not been able to give OUf attention arc thus brought into con-
formity with the Bohr principle.
Altogether we have in this very notable list of quantitative agree-
ments between experiment and the simple assumption of Bohr some
very strong reasons for believing that Bohr's model of the hydrogen
atom and the singly charged helium atom is something more than a
mere working hypothesis, and may represent an ultimate reality. Never-
theless we must be cautious in making such an inference, When
the theory of the luminiferous ether was in its prime. several models
or mechanical pictures of the ether were offered which represented
with some degree of adequacy the properties of the electromagnetic
field. We still recognize that the mathematical equations of hydro-
dynamics are largely identical with the mathematical equations of
clectromagnetics. but such mechanical pictures of the ether are now
thoroughly discredited.
It is to be remarked that while it seems natural to use the same
law of force between two charged parts of an atom which is found
to hold between two large charged bodies at greater distances from
one another, our satisfaction in the success of this experiment is some-
what diminished by the introduction of another assumption which
limits the validity of Coulomb's law to certain specific orbits, This
is especially true since no suggestion has yet been made regarding
the quantitative or even the qualitative laws governing the electron
between any two orbits. Indeed the combination of the quantum theory
with the Rutherford theory of the atom seems to result in a model
which has properties in some sense intermediate between those of
an atom whose parts are in rapid motion and those of a static atom,
as the fonowing considerations show.
It was predicted from the ~lassic~l electromagnetic theory that any
accelerated charge would emlt radIant energy, but the electron in
one of the stable orbits of the Bohr theory is subject to constant ac-
celeration toward the center of the atom and yet is supposed to emit
no radiation. However, such emission of energy from an accelerated
charge does not seem to be an entirely inevitable consequence even of
classical theory, and for this reason we may turn our attention to a
still more elementary way in which I have attempted to show (19 17)
the. divergence between the properties of the Bohr atom and the prop-
ertIes that would formerly have been assumed for a system containing
an electron in motion .
. Let us in Figu,re I I represent, a hydr~gen atom according to Bohr
wIth an electron In the first orbit, that lS to say in the most stable
state, and let us represent by AA' a small wire which may be brought
SPECTRAL SERIES AND PHYSICISTS VIEW OF THE ATOM 51
near to the hydrogen atom. Now if the electron in the orbit exerts
any sort of electrical force at a distance, when the electron is in position
X there will be a slight flow of positive electricity in the wire
toward A, and when the electron is at X' there will be a slight flow
toward A'. Indeed at any finite distance of the wire from the atom
there should be set up in the wire a finite alternating current which
would continue indefinitely. Such a current should generate heat,
but since the atom is supposed to he in the state of lowest possible
energy there appears to be no SOUTce from which the heat could
originate. In other words, we must conclude either that such an
A
FIG. H.-Illustrating a Doubtful Point Concerning the Bohr Atom.
alternating current is not produced or that it is produced but meets

with no ohmic resistance.
The latter alternative is possible. The electrons in the metal wire
may also be presumed to be subject to quantum laws, and since the
motion which they would acquire owing to their proximity to the
hydTogen atom would be very minuie, it might be assumed that their
displacement to and fro could occur without frictional loss. Indeed
it is perhaps to be predicted that with a perfect crystal at a very
low temperature a certain potential gradient will be found to be
necessary before the electrons can be sufficiently displaced from their
equilibrium positions to give ordinary electrical conduction. But with
ordinary metals at ordinary temperatures we have no experimental
evidence that Ohm's law would fail even at very small values of
the electromotive force, nor does this seem likely on theoretical
grounds.
If these considerations are correct, we must conclude that an elec-
tron, in a Bohr orbit, exerts upon other electrons no force which
depends upon its position in the orbit. In other words it seems as
though we should add another assumption to those of Bohr, namely,
that while the orbit of one electron may as a whole affect the orbit
of another electron, we should look for no effects which depend upon

the momentary position of any electron in its orbit. If this idea
proves useful it will greatly simplify attempts to secure adequate
models of atoms or molecules containing two or more electrons.
The remarkable quantitative success of the Bohr model was limited
to the special case of an atomic nucleus accompanied hy onc electron.
Efforts to constr11ct equally complete and adequate models of atoms
\vith two or morc electrons hayc so far failed. In Bohr's original.
tlwory of an atom containing a number of electrons, thbC electrons
were SUpfK.lSed to be arranged in successive concentric rings about
the central nucleus, the electrons in anyone ring being (,_'qually spaced
and moving with the same velocity. Howeyer, this idea has hcen
abandoned atHi in the next chapter we shall discllss the more recent
views of Bohr regarding the structure of such an atom.
Magnetic Phenomena.
Excepting the ohservations on spectral lines, there appears to be
no method of studying the structure of the atom which is so direct
or so promising as the method fl1rnish~d by the study of magnetism.
Unfortunately the experimental difficll1ties in this field are great,
and at present we hav~ vcry meagre data concerning tbe magnetic prop-
erties of substances. But even the information which we now pos-
sess is of the greatest importance to any theory of atomic and molecular
stntcture.
The behavior of suhstances in a magnetic field is in many respects
analogous to that in an electric field. \\Then two plates of an electric
condenser in a vacuum are oppositely charged they attract one another,
and if any ohject which has a positive charge at the one end. and a
negative charge at the other. 15 placed between the plates it tends
to orient itself so that its positive end approaches the negative plate
and its negative end approaches the positive plate. In so doing it
diminishes the attraction between the two plates in accordance with
the ordinary laws of electric attraction.
So also-when the space hetween the plates, formerly separated
by a vacuum, is filled by any substance the attraction diminishes,
and the ratio hetween the original attraction and the attraction now
observed is called the dielectric constant of the substance in question.
In such cases also it is assumed that the substance contains molecules
which are differently charged at the two ends and are therefore called
dipoles. The amount of motion of these charges, either through the
rotation or stretching of the dipole, or through slight displacements
of electrons or nuclei from the equilibrium positions which they
normally occupy, is believed to detenl1ine the magnitude of the dielec-
tric constant:
The molecular dipoles, which tend to orient themselves strictly in
the line of electric force, could not be expe(.'ted· to do so completely,
because of themlal agitation, and in accordance with this view it is
SPECTRAL SERIES AND PIIl"SICISTS I'lEW OF THE ATOM 53
found experimentally that the dielectric rQn:;tant of a substance always

diminishes with increasing temperature.
A ma&Tnet in a magnetic field behaves very much like the dipole
in the electric field. Its torque is proportional to what is called its
magl1etic l11Ol1U'1It. Tn the case of a simple har magnet this magnetic
moment is proportional to the ~ntensity of mag-netizatioll and the
distance between the two poles. If the magnet is an electric circuit
the moment depends upon the amonnt of current and the dimcnsions
of the circuit.
Two opposite magnetic poles, separated hy a vacuum, exert upon
onc another an attraction , .... hicll is diminished if a s111all mag-net is
placed hetween the two pules and I:; allowed to orient itself in con-
sequence of its magnetic torque, w that its south end approaches
the north Jlole of the magnet, and its north end approache .. :; the south
FIG. 12.-Illustrating thc Type of Orientation of a ),lokclIlar MaRnct That

ProduCts Paramagnetism.
pole (Figure 12). As in the analogous case of the electric field,

varions suhstances placed between the two magnetic poles diminish
the mutual attraction hetween these poles.
The ratio between the original attraction hetween the poles and
that which is observed after the intervening space has heen filled
with the substance in question, is known as the permeability of the
substance. \Vhen the attractive force is thus diminished, the sub-
stance is said to he paramagnetic, or in extreme cases, ferromagnetic.
It is assumed that the suhstance contains molecular magnets which
tend to orient themselves in the magnetic field, but that the thermal
agitation preYents complete orientation. It is an observed fad that
the permeability of such paramagnetic substances always diminishes
with increasing temperature.
There is, however, another class of substances, having no counter-
part in the electrical analogue, that increase the attraction uetween
two magnetic poles, and these substances which have permeability less
than unity are s::alled diamagnetic. In a typical diama!,'11etic substance
the penlleability is independent of the temperature.
In Langevin's theory of para- and diamagnetism, every molecule
contains within it electric circuits or orbital electrons. Each electric cir-
cuit is the equivalent of a small magnet, and these elementary circuits or
magnets 'are assumed not to be affected by temperature. nor are they
supposed to be free, within the molecule, to orient themselves ap-

preciably in a magnetic field. But if the several elementary magnets
which are contained within a molecule have a resultant magnetic
moment in one direction, then the molecule as a whole can turn in
the magnetic field and produce the phenomenon of paramagnetism.
\\lhile the individual electron orbits are not considered free to
orient themselves in a magnetic field, it is nevertheless shown to be a
consequence of electromagnetic theory that the fieJd wiJl produce a
slight change in the orbit itself, and this change is in such a direction
as to produce the phenomenon of diamagnetism. Thus according to
Langevin's theory every substance exhibits this diamagnetic phenome-
non, the effect of which, however, is overshadowed in paramagnetic
substances by the large effect of opposite sign which is due to the
orientation of the molecular magnets.
Quantitatively Langevin's theory permits a calculation of the dimen-
sions of the electron orbits which seem to be of the expected order of
magnitude, although it is now generally believed that the details of
his theory must be modified in accordance with quantum theory.
\Ve shall have occasion to return later to the consideration of the
magnetic properties of suhstances, but in the meantime it is evident
that, irrespective of any special theory, the mere existence of magnetism
in substances appears to imply that the atom contains charged particles
in rapid motion. It is conceivable that a stationary electron in a
position of asymmetric stress might give rise to a magnetic field. but
this suggestion has not hitherto been made, nor does it seem at pres-
ent capable of useful development. In the present state of science
it therefore seems best tu regard the existence of magnets as definite
evidence of electricity in motion.
Chapter IV.
Reconciliation of the Two Views; the
Arrangement of Electrons in
the Atom.
We have now discussed two distinct views of the internal structure
of the atom. According to both views the neutral atom is composed
of a central nucleus with a positive charge equal to Z, the atomic
number, and a group of Z electrons situated about this centeT.
The view based on the periodic law and the chemical behavior of
the elements leads to the picture of a relatively static atom. Accord-
ing to this picture, the electrons occupy fixed positions which are ar-
ranged in concentric shells about the nucleus. It is not imp1ied that
the electrons may not be displaced from these positions by the action
of heat and light, or driven into new positions when a chemical re-
action occurs, Nor is there anything in this view really incompatible
with the assumption of an electron in rapid motion, such as the Parson
ring electron, so long as the electron as a whole is regarded as
occupying a fixed position in the atom. This theory of the static atom
obviously abandoned the assumption that the ordinary laws of elec-
trical attraction and repulsion are valid within the atom.
The experiments of physicists led to a quite different view of the
atom. The theory of Rutherford assumes the forces between the
charged particles within the atom to be the same as those which hold
for massive charged bodies. The atom is regarded as a sort of plane-
tary system in which the force of attraction between the nucleus
and electrons is balanced by the centrifugal force due to their orbital
motion. The electrons are considered to be arranged in successive
rings rather than in successive shells.
These two views seemed to be quite incompatible, although it is
the same atom that is being investigated by chemist and by physicist. If
the electrons are to be regarded as taking an essential part in the proc-
ess of binding atom to atom in the molecule, it seemed impossible
that they could be actuated by the simple laws of force, and travelling
in the orbits, required by the planetary theory. The permanence of
atomic arrangements, even in very complex molecules, is one of the
most striking of chemical phenomena. Isomers maintain their identity
for years, often without the slightest appreciable transformation. An
organic molecule treated with powerful reagents often suffers radical
change in one part of the molecule while the remainder appears to
55
suffer 110 change. It appears inconceivable that these penllanent though

essentially un stahle configurations could result from the simple law
of force embodied in Coulomh's law.
The fIrst step toward removing barriers between the two types of
atomic model was made by Bohr when he restricted the application of
Coulomb's law to ~pecific states or orhits. 1 have attempted to show
in the preceding ehapter that it is the orbit as a whole rather than the
particular position of the electron within the orbit that is the thing
of essential interest in the Bohr theory. If these orhits are in fixed
;)(Jsitions and orientations the~' may be' used as the building stones of
an atom \\'hich has an essentially static character.
There remained, howe"er, ill Bohr's original theory some features
which were far from compatible with the chemist's view of the atom.
This is essentially true of his models of atoms containing more than
one dectron. Here he assumed rings of electrons, f(.'voiying in a com-
mon orhit. which seemed (Iuite irreconcilable with the common phe-
nom('n<1 of chemistry.
Also from the si<le of physics evidence began to accumulate which
\vas opposed to the ring theory. The X-ray spectrographs obtained
from crystals seemed to indicate a cubic or some other regular poly-
hedral structure of the electrollS alwot the atom, as was shown by the
investigations of Hull (1917). A like conclusion was reached hy
Born and Lande (] 9IR) in their searching physico-mathematical in-
\'('stigation of the ('ommon physical properties of crystalline substances.
These authors, \vhile maintaining the \'ie\\' of the orbital electron,
make the orhit small, and make the position of the orhit correspond
to the positions as:-ignec1 to the electrons hy Parson awl myself.
In his later work Bohr entirely abandoned the ring of electrons.
He fonnd that even the phenomena of spectral lines in the yisihle and
in the X-ray regio11s cOl1ld not be interpreted jJ1 terms of a theory
which regards the electrons as associated with one another in joint
orbits. 11 e now assigns to each electron its separate orbit and regards
these orbits as situated about the atomic center in shells.
It seems to me that by this step Bohr has removed every essential
element of conflict hetween the views of the physicist and the chemist.
If we regard as the important thing the (Jrbit as a whole, and not the
position of the electron within the orbit, and if each electron is as-
signed an independent orbit, then v·;e may think of each electron orbit
as having a fixed position in space. The average position of the
electron in the orbit may be called the position of the electron and
will correspond entirely to that fixed position which was assigned in
the theory of the static atom.
Let us therefore now attempt to weld these different views into a
single theory C!f atomic structure which. while it certainly can claim
no degree of '-finality, will summarize all of the evidence, chemical
amI physical, \yhich we now possess regarding atomic structure.
1. First we shal1 adopt the whole of Bohr's theory in so far as
it pertains to a single atom which possesses a single electron. There
RECONCILIATION OF THE TWO I'IEWS 57
arc no facts of chemistry which arc opposed to this part of the

theory. and we tints incorporate in the new model all of the Uohr theory
that is strictly quantitative.
2. In the case of systems containing more than onc nucleus or
more than one electron. \'Irc shall also assume that the electron possesses
orhital motion, for such motion seems to he required to acconnt for
the phenomenon of tl1agnetj~m; and each electron ill its orbital u1(Jtion
may be regarded as the equivalent of an elementary magnet or 111ag~
neton. Howcyer, in the case of these complex atoms and molccllles
we shall not assume that an atomic nucleus is necessarily the Ct'ntCf
or focus of the orbits.
3. These orbits occupy fixed po.-;itions \. . ith respect to one an-
other and to the nuclei. \Vhcn we speak of the position of an electron,
we shall refer to the position of the orbit as a whole rather than to
the position of the electt:on \vithin the orhit. \'lith thi~ interpretation,
we may state that th(' change of an electron from one position to an-
other is always accompanied by a finite change of energy. \Vhen the
positions are such that no change in position of the several parts of the
atom or molecule will set free energy, we may say that the system
is in the most stable state.
4. In a process, which consifts merely in the fall of an electron
from one position to another more stahle po"ition, monl)chromatic radi-
ant energy is emitted, and the frequency of this radiation multiplied
by It, the Planck constant, is equal to the difference in the energy
of the system hetwe(~n two states.
S. The electrons of an atom are arranged ahout the nucleus in con-
centric shells. The electrons of the outermost shell are spok('n of as
valence electrons. The valence shell of a free (uncombined) atom
never contains more than ciyht ciectrons. The remainder of the atom,
which includes the nucleus and the inner shells. is called the kernel.
In the case of the noble gases it is customary to consider that there is
no valence shell and that the wh(Jle atom is the kernel.
6. In my paper On "The Atom and the Molecule" I laid much stress
upon the phenomenon of the pairing of electrons. I have since hC{"(ltne
convinced that this phenomenon is of (Ten greater significance than I
then 'Supposed, and that it occurs not only in the yalenee shell hut also
within the kernel, and eyen in the interior of the nucleus itself. It has
not seemed des ira hIe to discuss in this hook the extremely interesting
modern ideas concerning the structure of the atomic nucleus, hut if we
adopt the old hypothesis of Prout it is possible from the atomic weight
and the atomic number alone to determine the number of hydrogen
nuclei and the number of electrons which compose the nucleus of a
given atom. It is a striking fact that with very few exceptions the
number of nuclear electrons so calculated is an even number. It is
furthermore to be noted that whenever a radioactive atom emits one
beta-particle it almost immediately emits another, again illustrating the
instability of an unpaired electron within the nucleus. So also we
find that in all the more stable states which atoms assume, the
electrons occur in even numbers in the several inner shells. Later we

shall show that the valence electrons almost invariably follow the same
rule. The simplest explanation of these facts appears to lie in the
assumption of a physical pairing of the electrons. There is nothing
in the known laws of electric force, nOT is there anything in the quantum
theory of atomic structure, as far as it has yet been developed, to
account for such pairing. However, we have seen that an electron
within the atom must be regarded as a magnet, and two such magnets
would tend to be drawn together. \Vhile the classical theory of mag-
netism would hardly suffice to account fully for this phenomenon of
pairing, there can be no question that the coupling of electrons is inti-
mately connected with the magnetic properties of the electron orbits,
and the explanation of this phenomenon must be regarded as one of
the most important outstanding problems in quantum theory.
7. We may next consider a very recent, idea advanced by Bohr
(1921), which is not based so much upon deductions from his atomic
model as upon a direct consideration of the experimental data on spec-
tral series. He assumes essentially that the first shell is associated with
a single energy level, and that this level can accommodate one pair of
electrons, that the second shell contains two energy levels, each of which
is capable of holding two pairs of electrons, making a maximum of eight
electrons in the second shell. The third shel1 has three energy levels,
each of which can hold three pairs of electrons, so that the maximum
number of electrons in the third shell is eighteen. The fourth shell
comprises four levels, each capable of holding four electron pairs, mak-
ing a total of thirty-two electrons, and so on. \Ve shall see the great
utility of this conception as we now proceed to consider the arrange-
ment of electrons in the various elements.
The Inner Structure of the Several Atoms.

In hydrogen the kernel of the atom is the nucleus itself, and there
is one valence electron. This single electron should give to the atom
a large magnetic moment, and we expect to find that monatomic hydro-
gen is highly paramagnetic. Unfortunately no one has succeeded in
devising an experimental method of ascertaining the susceptibility of
the monatomic form, which can only be obtained at very high tempera~
ture, or through the agency of a powerful electric discharge.
In a recent investigation of extraordinary interest Stern and Gerlach
(1921) have succeeded in studying in a most direct manner the magnetic
properties of the silver atom which, like the hydrogen atom, we suppose
to have one electron in its outer shell. They find that its behavior is
similar to that which would be predicted for the hydrogen atom by the
Bohr theory. ,
(There is of course a possibility that the nucleus of an atom might
itself possess a magnetic moment. As a rule, the magnetic properties
of substances are found to be largely dependent upon the physical state
of an element or its state of chemical combination, so that it may be
RECONCILIATION OF THE TWO VIEWS 59
assumed that the outer electrons, which are the ones chiefly affected in
ordinary physicaJ and chemical changes, are the ones that are respon-
sible for magnetic properties. Nevertheless we have no direct evidence
that the nucleus may not itself be magnetic.)
In diatomic hydrogen and in helium it was originally assumed by
Bohr that the two electrons revah'cd in the same direction in the same
orbit. This would produce a large magnetic moment, and the two
gases would therefore be expected to be param~gnetic. On the con-
trary both gases are diamagnetic, and this model is evidently unsatis-
factory. A later model in which the two electrons are placed in sepa-
rate orbits, making an angle with one another, may be criticized on the
same grounds. since this arrangement also should lead to paramag-
netism. There are two ways in whjch a pah of magnets can be heJd
..s N
ttN S
FIG. 13.-Two Ways in Which a Pair of Magnets May Be Drawn Together.
together by their magnetic forces. These are illustrated in Figure 13.

in which the magnets are represented either as electric circuits or as
the equivalent bar magnets. Jn the first arrangement the two magnets
augment one another and produce a magnetic moment greater than that
of either magnet alone. In the second arrangement the two magnetic
moments neutralize one another. and the lines of magnetic force are
almost entirely confined to the immediate neighborhood of the pair of
magnt;ts. Since the phenomenon of diamagnetism is predominant in
the great majority of substances, we must assume that the second
arrangement more nearly represents the normal condition of an elec-
tron pair.
The two electrons of helium are not ordinarily regarded as valence
electrons, and therefore we may consider the whole atom of helium as
the kernel. In paragraph 7 of the preceding section we have seen that
Bohr postulated that the first shell of an atom corresponds to a single
energy level and can accommodate two electrons. The pair of electrons
in the helium atom constitutes this shell. and a similar pair will appear
as the first shell of all atoms of higher atomic number.
The system of nucleus, and first shell of two electrons, forms tne
kernel of the elements from lithium to fluorine inclusive. The neutral
atom of lithium has a nuclear charge of + 3. and one valence electron,
(Xl FALl'NCl, AND TIll, STRUCTURE OF .1TOMS AND MOLECULES
which is readily given up I)y the atom. 1eaying the kernel, which is the
lithium ion. Bervllium has two valence electrons, which are also easily
lost, gi\'ing the 1)cryllium ion. On the other hand, boron with three
valence electrons does not so readily lose these electrons, and the free
horon ion is not known to exist. S~) carbon has four valence electrons,
nitrogen ti\'C, oxygen six, and fluorine seven. In the last three cases
there is an increasing tendency to take up additional electrons to form
the stahle group of eight, Of the octet, which we have shown in Chapter
II to be one of the most striking phenomena of chemistry. The nitride
ion with three extra electrons probably exists, although this has not
heen definitely proved. The oxide ion possesses two extra electrons,
while fluorine reacts violent!" with almost all substances in such manner
as to take up the one electro;1 necessary to form fluoride ion.
In neon the neutral atom contains this second shell of eight electrons,
and forms an extremely stahle system. Here again. as in the case of
helium, we may consider the kernel to comprise the whole atom, namely,
the nucleus) the first shell of two, and the second shell of eight. In
terms of the Bohr assumption. the outer group of ejght electrons repre-
sents the maximum number of electrons which can enter the second
shelL According to his idea. four of these electrons are on a somewhat
different energy level than the other fonr, but thi,c; is a distinction that
we may disregard when considering the purely chemical properties of
the atoms.
A kernel entirely resemhling the neon atom characterizes all of the
clements of the next period. \Ve may represent the distribution of
electrons in the successive shells in the following manner: Na, 2-8-1;
A1g, 2-8-2; and so on tn CI. in which the distribution is 2-8-7. In each
case the last figure gives the number of valence electrons, and, as in
the preceding period, the elements with a small number, of valence
electr(111S tend to give these up to form positive ions. \\rhile those with
a greater number t('no to take up enough electrons to form negative ions
with a complete group of eight in the outer shell.
The First Long Period.

The distribution of electrons in argon may be represented by 2-8-8,
and by analogy tn the two preceding periods we might have expected
a kernel of this type to characterize the elements of this period, and
indeed we find that the chemical properties of the first three elements
are in accordance with the arrangements: K, 2-8-8- T; Ca, 2-8-8-2;
Sc, 2-8-8-3. But when we come to titanium we see this simple system
breaking down. Indeed we have postulated that no atom ever has
more than eight valence electrons, and we are dealing with a period
of eighteen elements, so that the series cannot proceed from the first
to the last element merely by the addition of successive electrons to the
outer shell. It is evident that some new phenomenon. must occur.
1\1any years ago I was led to the conclusion t1k1.t when we speak
of an element of variable valence we are including two very different
RECONCILIATiON OF THE TIVO VIEIVS til
types of phenomena, and that such a change as the oxidation of

hydrogen sulfide .to sulfurous or sulfuric acid is a very different kind
of change frotTI the one which occurs when a ferrolls salt is converted
into a ferric, or a titanous salt into a titanic. There afC numerous
reasons for suspecting a fundamental difference between these two
types of oxidation. Fur example, in the first type oxidation and reduc-
tion usually occur by steps of two, \vhile the senmd typ<: mOfe often
invoh'cs a change of but one step. Suhstances which undergo the
second type of change afe usually colored, while thns(' invohTd in the
first type of change afe usually colorless. The distinction between the
two types was expressed in my paper of 11)16 by referring to the atoms
which undergo the second type of valence-change as atoms of 1'aria{;/c
/wnlcl.
The properties of the yarious elements which come hefnre titanium
are in accord with the assumption that the elcctrons 'which are a part
of the kernel do nut take part in chemical l·cactions. hut as we now
proceed to elements of higher atomic number we shall sec that this
simple rule is by no means uni \'crsally valid.
It is a remarkahle fact concerning the metals that we have so far
discussed that when they form ions they give off simultaneously all of
the electrons of the outer shell. \Yhcn Ca acts as an electrode in
electrolysis, its atom never loses a single electron to form the ion Ca+.
If it docs, this ion must be unstable and react immediately according to
the reaction. 2Ca+ = Ca + Ca~+. It is true that in the ionization of cal-
cium vapor the ion Ca" is assumed to account for tlw so-called enhanced
spectrum of calcium, but presumably two such ions coming together
would undergo the above reaction. Compounds of the type Cae! arc
unknown. Aluminum docs 11'.)t form the ions Al+ and AI"+, nor arc
compounds known of the type AICI and AICt. In other words, when
such a metal reacts it uses all of its \":lIenee electrons or none. (It is
probable that in some atoms ()f high atomic weight there arc excep-
tions to this rule. It seems likely that in the change from thallnus
to thallie ion ur from aurous to auric ion there is no change in the
kerne1.)
Now let us see the consequence of applying this rule to the elements
in the periodic table which we have now reached. Titanium gives three
classes of compounds, one corrcsponding to the titanous ion, Ti++, one
to titanic ion, Ti+++, and one presumably to the quadripositive ion,
Ti++++. Now if we are to believe that in each of these ions the atom has
given up all its valence electrons, then We must assume that there are
three possible states of the titanium atom containing, respectively, two,
three, and four valence electrons. Since the total number of electrons
must be the same for any state of the neutral atom, this means that
the kernel must have one more electron in the first state than in the
second, and one more in the second than in the third. If we assume
that the extra electrons go into the preceding shell, we can represent
the three states of the neutral atom as 2-8-8-4 (quadripositive), 2-8-9-3
(tripositive, titanic), and 2-8-10-2 (bipositive, titanous).
6, VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULE
TABLE.
COMPOSITION OF ATOMIC KERNELS IN ELEMENTS OF THE FIRST LONG PERIOD.
2-8--8 2--S-<; 2-8-10 2--&-11 2-8-1:1 2-8-13 2--8--14 2+15 2-8-16 2-8-17 2-8-1
A
K<
Ca++
5c+++
Tj++++ Tj+++ Tjt+
Cr t +';' Cr t ++ Cr-++
.;.++ +++
Mn HH i\1n+-H
Fe+++ Fe+++ F~++

Co<-++ CO"""
Ni H + NiH
Cu- Cu'"
Zn H
Ga+++
etc.
It wiJI be understood that in writing such formula- as Crt+ and Mn++

we are not necessarily assuming the existence of these ions, but simply using
shorthand m{'thod of expressing the state of chromium in chromates and (
manganese in permanganates.
RECONCIUATiON OF THE TWO VIEWS
Applying the same idea to the succeeding members of this period

of elements of variable kernel, we find the distrihution of electrons in
the several kernels as shown in the accompanying table. Thus the
kernel represented by 2-8-8 is the only one that occurs in argon, potas-
sium, calcium and scandium. It also occurs in titanates, vanadates,
chromates and permanganates. Every form of kenlel from 2-8-8 to
2-8-18 occurs in the table. Thus the kernel 2-8-II is contained in
vanadous compounds, in chromic compounds, and in manganese dioxide.
When we come to the kernel 2-8-18 we find evidence that this com-
position of the kernel, like the form 2-8-8, appears to possess a higher
degree of stability than the transition forms. Thus we find the kernel
of copper assuming this form in the cuprous compounds, and all the
remaining elements of the period, Zn, Ga. Ge, As, Se and Br, adopt
no other form of kernel than 2-8- 18.
It is evident that some at least of the electrons in the kernels of
the various transition forms between 2-8-8 and 2-8- I 8 are in a very
different state from that which characterizes the electrons of the inner
shells of the elements which we have considered hitherto. Simple
processes of oxidation or reduction suffice to diminish or to increase
the number of electrons in the kerneL That some of these electrons
are not tightly held is shown also in the prevalence of color in the
compounds of these transition elements. An equally striking charac-
teristic of the transition elements is found in their magnetic properties.
To say nothing of the remarkable ferromagnetic properties of the
elements themselves, we find that many of the compounds of these
metals are strongly paramagnetic.
Indeed we find that pronounced paramagnetism is almost exclusively
the property of these elements and of the transition elements of the
remaining long periods in which the phenomenon of variable valence
also appears. Thus if we inspect the very incomplete data given in
the tables of Landolt, Dornstein and Roth, we are struck by the fact
that (with the remarkable exception of molecular oxygen, O 2 ) there
is listed no elementary or compound substance possessing a higher
magnetic susceptibility per gram than 10-6 , except those containing
titanium, vanadium, chromium, manganese, iron, cobalt, nickel and
cupric copper in the first long period; columbium, rhodium and pal-
ladium, which are transition elements in the second long period, and
cerium and praseodymium in the third long period.
It is moreover to be noted that the highly paramagnetic character
of these elements disappears when the kernel assumes either of the two
forms 2-8-8 and 2-8-18. Thus chromates and permanganates are
weakly paramagnetic, while cuprous compounds are either slightly para-
magnetic or diamagnetic. On the other hand, the most highly magnetic
substances are those which occur in the center of the transition group.
In this transition stage which occurs before the completion of the
new kernel-and this same process apparently occurs in each of the
succeeding periods of the elements-there is a marked departure from
the rule which is so generally observed, that electron shells tend to
hold an even number of electrons. If we' consider the bivalent ions

of titanium, vanadium, chromium, manganese, iron, coba1t, nickel,
copper, there is no alternation of stability between the forms with
even and odd numbers of electrons in the kernel.
While in a stable kernel the electrons, regarded as elementary electric
circuits or magnets, seem to be arranged in such manner as to neutralize
the magnetic fields and to prevent a resultant magnetic moment, it
seems to be quite otherwise in the case of these unstable forms of kernel
in the transition group. Without attempting to suggest the actual
structure of these unstable kernels, it is perhaps interesting to note that
the magnetic properties which have been mentioned would result if,
after the kernel has the form 2-8-8, the additional electron orbits (re-
garded as elementary magnets) entered into the kernel in' a row, the
north end of one djrectly toward the south end of the next, so that
each additional magnet would increase the total magnetic moment. If
S--N "
"
./~ \
\
\
\
R:
1
~\ ,
,;'
" " ......-
FIG. 14.-A Possible Arrangement of Elementary Magnets in the Transition
Metals.
instead of a row we should imagine a ring of ten magnets, as ,in Figure

14, this ring being built up one magnet at a time, then at the beginning
each magnet would increase the resultant magnetic moment, but as the
ring approached completion the moment would again diminish, and in
the completed ring the magnetic moment would disappear. This may
give a crude picture of the way in which the stable and non-magnetic
kernel 2-8-8 passes into the stable and non-magnetic kernel 2-8-18, with
the highest magnetic moment in the middle of the transition period.
If a metallic ion is merely a kernel which has lost all its valence
electrons, and if the structure of this kernel determines the magnetic
properties of the ion, then (assuming that the magnetic properties of
the ions of the iron group are not much affected or are equally affected
by hydration~ we might expect a dose resemblance between two ions
which have precisely the same type of kernel and differ only in the
charge of the atomic nucleus. Kossel has called a:ttention to some very
remarkable results obtained by \Veher (1915) on the atomic suscep-
tibility of such ions. It appears that the paramagnetic susceptibility
RECONCILIATION OF THE TWO VIEWS
of the bh'alent ions increases from the chromous to the manganous ion
and then diminishes to the nickelous ion. He obtained the following
figures;
Ion CrH 1fn++ Fe+ COH NiH
"I.. 0.011 0,015 0.01 3 0.010 0.004
Studying the corresponding trivalent ions, \'Vcber finds Xu. == 0.006 for
chromic ion. Manganic ion, which has the same kind of kernel as the
chromous ion, has also the same susceptibility, namely, O.OIl. Ferric
ion, with the same type of kernel as the manganous ion, has its sus-
ceptibility, namely, 0.015. (Cobaltic ion was found to ha-ve a much
lower susceptibility than ferrous ion, but this was very like1y due to
the difficulty of preparing a pure coLaltic salt.) Such facts show very
dearly that in these ions it is the type of kernel that determines the
magnetic properties, and that the magnetic susceptibility is at a maxi-
mum in the kernel which has five more electrons than are present in
the kernel 0f argon.
All of these facts seem to wa.rrant the assertion that in this first
long period we are dealing with a group of elements many of which
show the phenomenon of a variable kernel; and that in proceeding
through the successive elements from scandium to zinc, ten electrons
have been added to the kernel. This is entirely in accordance with
]3ohr's rule. \Ve have seen that the kernel of the helium type has its
first shell fully occupied by the pair of electrons which alone can be
accommodate,l at this energy level. In the kernel of the neon type
the second shell also contains its full quota of eight electrons, but in
the kernel of the argon type, 2-8-8, the third shell contains only eight
of the eighteen which that shell is capable of holding. Therefore we
may conclude that, in passing through the transition group from scan-
dium to zinc, the ten electrons \vhich are added to the kernel comp1ete
the third shell of eighteen electrons, so that in krypton we have a
kernel of the type 2-8-18-8.
The Remaining Periods.

The fourth atomic shen should be capable of holding thirty-two
electrons, and we might expect the next long period, beginning with
rubidium, to fill all these places. In fact, however, we find an almost
exact duplication of the preceding period. Once more we pass through
a series of transition elements, columbium being the first to show a
variable kernel. Each of the four preceding elements, rubidium, stron-
tium, yttrium and zirconium, gives but a single ion, and the compounds
of these elements are colorless and have a Jow magnetic susceptibility.
The succeeding elements behave very much like those of the first long
period. and the final element xenon may be assigned the kernel structure,
2-8-18-18-8.
This kernel of the xenon type persists in the first members of the
next period, namely, cesium, barium and lanthanum. In cerium we
come to the first element of variable kernel, and now we may assume
that the electrons which are added to the kernel complete the fourth
shell of thirty-two electrons. Such a process would involve no change
in the valence shell from element to element, nor in the next inner
shell, and would thus produce only subterranean changes in the atom.
Thus we may account for the great similarity in the properties of the
rare earths.
The transition process may be assumed to be completed when the
fourth shell has its full complement of thirty-two electrons. and the
fifth shell eighteen of its possible fifty, so that the kernel of the noble
gas, niton, which concludes this period, may be considered to have the
structure 2-8-18-32-I8-8.
This is essentially the view of the composition of the inner shells
which was proposed almost simultaneously by Bohr (1921). who con-
sidered the ::;pectroscopic behavior of the elements, and by Bury (1921)
who considered their physico-chemical properties. It possesses much
verisimilitude and furnishes unquestionably the best picture of the
atomic kernel which our existing knowledge permits. Some thlngs
which still seem obscure will doubtless be clarified by further study of
the ordinary and X-ray spectra of the elements, and of their magnetic
properties and ionizing potentials.
In the meantime it must be admitted that the prohlem is by no means
completely solved. Both Bohr and Bury assume that the last frag-
mentary period of the Mendeleeff table is a period of thirty-two ele-
ments, but I have pointed out in a previous chapter that the properties
of thorium and uranium indicate a far less resemblance to the preceding
period of thirty-two than to the next preceding period of eighteen.
Then again the cause of the stability of certain groups is not yet
adequately explained. Presumably the relative energies of the several
possible states will determine whether an electron will appear in the
outer shell of an atom or enter one of the inner unfilled shells, but while
we may grant that the numbers 2, 8, and 18 represent the maximum
number of electrons which may enter the first, the second and the third
shel1s of the atom, we have no indication as to why these same groups ,
appear in unfilled shells. We still have no explanation for the ex-
tremely significant fact that all of the rare gases (except helIum) and
all the common elementary ions, such as S--, Cl-. K+, Batt and Al+-l-t,
contain just eight electrons in the outermost shell.
Neither have we any adequate explanation for the pairing of elec-
trons which seems to characterize all the stable configurations of the
kernel. The tendency to form pairs and the tendency to form groups
of eight we shall find to be the essential features in the arrangement
of valence electrons in compound molecules.
Chapter V.
The Union of Atoms; the Modern
Dualistic Theory.
\Ve have now obtained a view of the structure of the atom which,
although perhaps incomplete or even erroneous in some of its details,
may safely be regarded as giving us in the main an adequate idea of the
arrangement of electrons about the atomic nucleus. OUT next task is
to ascertain if possible the way in which two or more atoms combine
with one another, and thus to obtain a similar intimate picture of the
structure of the molecule. In short we must endeavor to interpret the
great body of fact and hypothesis comprised in what is commonly
known as valence theory.
As I have remarked in another place (I9I3), "There is always the
danger in scientific work that some word or phrase will be used by
different authors to express so many ideas and surmises that, unless
redefined, it loses all real significance. Thus the term valence has been
used in discussing a large number of ideas which have perhaps nothing
more in common than the acceptance of Dalton's law of multiple pro-
portions. Even the conception of valence as an integral number has
been abandoned by those who speak of 'partial valence.' "
However, to those who have been responsible for the great achieve-
ments of structural organic chemistry, the idea of valence has not been
amhil:,'Uous. The valence of an atom in an organic molecule represents
the number of bonds which tie this atom to other atoms. Moreover in
the mind of the organic chemist the chemical bond is no mere abstrac-
tion; it is a definite physical reality, a something which binds atom to
atom. Although the nature of such a tie remained mysterious, yet the
hypothesis of the bond was amply justified by the signal adeqnacy of
the simple theory of molecular structure to which it gave rise.
The great success of structural organic chemistry led to attempts
to treat inorganic compounds in a similar manner, not always happily.
I still have poignant remembrance of the distress which I and many
others suffered some thirty years ago in a class in elementary chemistry,
where we were obliged to memorize structural formulce of a great
number of inorganic compounds. Even such substances as the ferri-
-cyanides and ferrocyanides were forced into the system, and bonds were
drawn between the several atoms to comply with certain artificial rules,
regardless of all chemical evidence, Such formula: are now believed
to be almost, if not entirely, devoid of scientific significance.
67
Such abuse of the structural formula inevitably led to a reaction

which found its best expression in the publications of Werner. His
"N euere Anschauungen auf clem Gebiete cler anorganischen Chemie"
(190S) marked a new epoch in chemistry; and in attempting to c1arjfy
the fundamental ideas of valence, there is no work to which I feel so
much personal indebtedness as to this of \Verner's. While some of
his theoretical conclusions have not proved convincing, he marshalled
in a masterly manner a great array of facts which showed the incon-
gruities into which chemists had been led by the existing structural
formulce of inorganic chemistry.
Especially \Verner called attention to the almost complete analogy
between the union of an anhydride with water to give the oxygen
acids, and the union of many halides with hydrogen halides to form the
halogen acids. Thus the reaction .f-r~O +50 3 == H~S04 is analogous
to the reactions HF + Ill', =HEI', and 2HCl +
PtCI, =
H,l'tCI,.
Now sulfuric acid could be giycn the structural formula
o
II
lI-O-S-O-H
II
o
but no such bonded structure could be given for the two halogen acids.
Therefore the latter were said to be molccular compounds. However,
there is no such fundamental chemical difference bctween the two types
as would warrant such a distinction, and I believe that anyone who
reads all of the facts as assembled by \Verner must admit that we must
either assign to hydrotluoboric acid a structure not very unlike that of
sulfuric, or dispense with structural formula! altogether.
The growth of such ideas and the success of the electrolytic dis-
sociation theory of Arrhenius once more drew the attention of chemists
away from the structural formula and directed it to the electric state
of the atoms of a molecule. Thus there came into being what may be
called the modern dualistic hypothesis.
The older dualistic theory developed by Davy, and especially by
llerzelius, was abandoned largely because it was discovered that the
positive element, hydrogen, could be replaced in organic compounds
by the negative element, chlorine, without producing any great change
in the physical properties of the compound. .But later measurements
of electrical conductivity showed that, after all, acetic acid and trichlor~
acetic acid are in some respects very different types, and that these
differences are readily interpreted on the old assumption that chlorine
has a far greater attraction for negative electricity than hydrogen has.
In the theory of Berze1ius, electricity was regarded as a fluid which
could flow to a greater or less extent from one atom to another, but
the recognition of the atomic nature of electricity made it now seem
necessary to assume that if the molecule is made· up of charged atoms
each atom must have an excess or a deficiency of an integral number
THE UNION OF ATOMS; THE MODERN DUALISTIC THEORY ti9
of electrons, as compared with the neutral state. Thus the modern
dualistic theory regards chemical action as primarily due to the jump-
ing of electrons from atom to atom.
The properties of electrolytes certainly point to such a separation
of charges. If sodium and chlorine combine to form aqueous sodium
chloride, the sodium atom definitely loses and the chlorine atom gains
an electron. It seemed not unnatural to suppose that each molecule of
undissociated sodium chloride might also contain the same charged
particles, namely, sodium with one electron missing, and chlorine with
one additional electron, held together by electric forces. The same
theory might thea be applied to hydrogen chloride, and then to water,
and alcohol, and perhaps eventually to substances like methane and
hydrogen.
\Vhile the number of different ions which have been shown to exist
at high concentrations in aqueous solution is limited, there are many
other ions that can be shown to be present in small amount, and there
are others which may be assumed to exist in minute amount in order
to explain the behavior of various substances. Thus in a solution of
sodium aluminate there is an extremely small but perfectly calculable
concentration of aluminum ion. It might therefore seem reason-
able to regard each aluminum atom, even in the aluminates, as in
the same electrical condition, namely, with a positive charge of three
units.
In ~ome similar cases the ions have not even been proved to exist,
but the hypothetical ion, Cr:::, might he considered to represent the
actual state of the chromium ato111 in chromic anhydride or the chro-
mates. By the same method the electrical state of the sulfur atom would
be represented by S:!! in sulfates, by S+H+ in sulfites, and by 'S-~ in
sulfides.
Now such a theory obviously gives a very simple interpretation of
those processes in which a given element is said to be oxidized or
reduced by a certain number of steps. Thus if manganous ion is
oxidized to permanganate ion, it is said that manganese undergoes five
units of oxidation, and some other atom or atoms must at the same
time be reduced by five units. In the modern dualistic theory this
simply means that the manganese atom changes in state from a defi-
ciency of two electrons to a deficiency of seven, that is, from Mn++ to
Mn++::+, and that five electrons are taken up by other atoms.
In aU chemistry there is no concept which is more fundamental
than this one of reduction and oxidation. Under the name of phlogis-
tication and dephlogistication such processes were recognized even
before the discovery of oxygen. and this mode of classifying chemical
phenomena has ever since been regarded as one of the greatest utiJity.
Noone can doubt the desirability of placing in a class by themselves
a~m<:mia, !llkyl am~nes and am?1~nium sal~s, i~ another class nitrogen
tnoxlde, nttrous aCid, and the mtntes, and tn stIll another class nitrogen
pentoxide, nitric acid, and the nitrates. But the question remains as
to whether such a classification is an absolute one.
Convenient, and indeed necessary, as it is in the study of many
chemical reactions, can this classification be applied without ambiguity
to all chemical substances? If it can, it means that we may assign
every atom in every compound to a definite oxidation-reduction stage,
and this in turn, according to the electrical interpretation, implies a
complete acceptance of the modern dualistic theory which ascribes to
each atom in a compound an integral number of units of positive or
negative charge.
This number, positive or negative (or zero), intended to show the
electrical state of each atom, has sometimes been called the positive or
negative valence of the atom, but since these terms have also been used
to represent the maximum positive or negative charge which the ele-
ment can assume in a great variety of compounds, rather than the actual
charge of the atom in a given compuund, Bray and Branch (r913)
proposed the better expression, polar number. Thus in ferrous sul-
fate, FeSO", iron is assigned the polar number + 2, sulfur the polar
number + 6, and each oxygen - 2.
It is customary in works on inorganic chemistry to call ferric iron
trivalent and ferrous iron bivalent. Now this phraseology is not only
out of harmony with that of organic chemistry, but it is essentially
bad in that it gives a number without a sign. Thus nitrogen would
be said to be trivalent both in ammonia and in nitrous acid. In all
cases where the polar number is to be indicated I propose a slightly
different mode of expression. Let us say that nitrogen in ammonia
is t1'i41cgativ(', and that it is tripositivc in nitrous acid. So also let us
say that iron is tripositive in ferric salts and bipositive in ferrous salts.
It is to be noted that these lX)iar numbers may be convenient in a
systematic treatment of chemical reactions, even if there is not that
definite distriiJution of atomic charges required by the dualistic theory.
To quote from my paper (1913) which was published with that of
Bray and Branch, "Oxidation of an element means an increase of its
polar number, reduction means a decrease, and this simple system
furnishes an adequate method of dealing with all cases of oxidation
and reduction. It must be remarked, however, that on account of its
very generality this system would apply equally well even if purely
fanciful values of the polar number were chosen, provided that the
rules required by the fundamental law of the conservation of electricity
be observed. Moreover, non-polar compounds may be treated provi-
sionally as polar, and fictitious polar numbers may be assigned without
leading to any false conclusions."
Let us consider a very interesting case cited by Bray and Branch.
The substanq:.~ C OH S S0 2 0H might be regarded as a derivative either
of sulfuric or of sulfurous acid, namely, (a) sulfuric acid in which one
hydroxyl group is replaced by phenyl, (b) sulfurous acid in which one
hydrogen is replaced by phenyl. In accordance with assumption (a)
we should assign to sulfur the polar number + 6 and to phenyl as a
THE UNION OF ATOMS; THE MODERN DUALISTIC THEORY 71
whole the polar number - I. With assumption (b) we should give

+ 4 as the polar number of sulfur and + 1 as that of phenyl. Now
it is taken for granted by the dualists that on hydrolysis a positive fadi·
cal adds hydroxyl, and that a negative radical adds hydrogen. But
when we apply this criterion to the case before us, we find that the
substance hydrolyzes in acid solution to give benzene and sulfuric acid,
but in a basic medium it gives phenol and sulfurous acid. It was
shown by Bray and Branch that in order to preserve intact the idea
that every atom must have a definite polar number. it would be neces-
sary to assume the existence of two distinct tautomers in equilibrium
with each other, namely, one with the polar numbers S = + 4 and
C,H, = + I, and the other with S = + 6 and C,H, = - I.
However, in the papers by Bray and Branch and by myself which
have just been cited, it was very definitely shown that many compounds
give no evidence of that large separation of electric charge which is
required by the modern dualistic theory, and indeed that there are many
substances, of the type which we call non-polar, whose properties in-
dicate little or no electric displacement within the molecule. The
properties of a substance like sodium chloride differ SO radically from
those of a substance like molecular hydrogen that it seemed necessary
to distinguish in degree if nDt in kind between the extremely polar and
the relatively nDn-polar types, although I remarked that "It must not be
assumed that anyone compound corresponds whoJIy and at all times
to either type."
The same idea was very forcefully advanced by J. J. Thomson
(1914) who, in addition to a large amount of chemical evidence, ad-
duced physical evidence based on experiments with positive rays and
measurements of dielectric constant, to show the great difference
between different substances as regards the degree of polarization, or,
as he preferred to call it, "intramolecular ionization." This was a
radical departure from Thomson's original view in which, in accord-
ance with the views of Abegg, he assumed aU chemical action to be due
to the transference of electrons from one atom to another. This earlier
view had given impetus to an attempt to apply the full theory of electro-
chemical dualism to organic compounds. Numerous papers by Falk
and Nelson (1910), Fry (19Il), and others developed the "electron
conception of valence," and this theory still has vogue among many
organic chemists, as may be seen from a recent interesting paper of
Stieglitz (1922). It is to be borne in mind that the assumption em-
ployed by these authors is no other than the assumption made by many
inorganic chemists that each atom in a molecule exists in a definite
oxidation-reduction stage which can be represented by an integral
polar number.
Falk and Nelson began their work by quoting from Thomson as
follows: "For each valency bond established between two atoms the
transference of one (negatively charged) corpuscle from the one atom
to the other has taken place, the atom receiving the corpuscle acquiring
a unit charge of negative electricity, the other by the loss of a corpuscle
72 VALENCE AND THE STRUCTURE OF ATOMS-AND MOLECULES
acquiring a unit charge of positive. This electrical process may be

represented by the producing of a unit tube of electric force between
the two atoms, the tulle starting from the positive and ending on the
negative atom . . . . There is, however, one important difference be-
tween the lines representing the bonds and the tubes of electric force.
The lines used by the chemist are not supposed to have direction . . . .
On the electrical theory, however, the tubes of electric force are re-
garded as having direction starting from the positive and ending on the
negative atom . . . ."
In order to express in the chemical formula the electrical transfer
from one atom to another, Falk and Nelson drew an arrow pointing
from the atom which had lost an electron to the atom which had gained
one. Thus methane, in which each hydrogen atom was assumed to have
a positive charge of I and the carLon atom to haye a charge of - 4;
and carbon tetrachloride, in which each chlorine atom was assigned a
charge of - 1 and the carbon atom a charge of +
4, were repre-
sented by
H CI
.t, t
H--->C<.--II CI<.--C--->CI
t .{.
H Cl
So methyl alcohol would he represented as
H
t
H--->C___,O<c-n
t
H
It is evident that the arrows which represented the transfer of an
electron were now being used as a substitute for the traditional organic
bond.
However, there are numerous objections to such a complete appli-
cation of the dualistic theory to organic chemistry. Only a few of
these need be mentioned.
If the negatively charged chlorine atom in carbon tetrachloride is
identical with the chloride ion which comes from electrolytes such
as sodium chloride, it might be expected to exert a field of force
extending in all directions, and to attract any positively charged
atom, but there is nothing in the physical or chemical behavior of
carbon tetrachloride to indicate such a state of affairs. Again, the
arrow used by Falk and Nelson cannot represent a chemical bond
unless it means something more than was stated by the authors (and
by Thomson), for as far as they state, the formula for methyl
chloride
THE UNION OF ATOMS; TIlE MODERN !JU.1USTIC THEORY 73
H
~
H --+ C --+ Cl
t
H
would merely indicate that each hydrogen has a positiYe charge. that
the chlorine has a negative charge. and that the carbon atom has a
charge of - 2. But there is nothing in all this to show from which
" particular atom a given electron came. \Vc do not think of an electron
leaving a trail behind it, as a spider wea\'cs its web, but if Bot, wlmt is
the Si,h>11ificance of an arrow?
One of the formula:: proposed by Falk and Nelson for ethylene is
H-->C--+C~H
H --+<-- <-- H
They recognize the logical difficulty invoh'ed in such a formula when
they write, "In those cases where one valence proceeds in one direc-
tion and one in the other, it is assumed that the corpuscles whirh are
transferred arc localized on the atoms, as otherwise the carbon atoms
would become electrically neutral." Dut even so we might as well
indicate that a carbon atom has lost an electron in one part and g'd.ined
an electron in another part Ly vertical arrows, as follows:
tC<--H
I <-- H
Here, however, there is left no indication of a bond holding the two
carbon atoms together.
If we adopt the dualistic theory and regard the organic molecule
as an assemblage of charged atoms, it would seem impossible to ac-
count for the complex structures which often remain unchanged for
long periods of time. \Ve should rather expect such a structure to
tumble into the particular aggregations which would represent the
minimum of electrical energy.
Finally it may be noted that the dualistic theory led to a prediction
which has received no verification. W. A. Noyes (see Noyes and
Lyon, 1901), who very eady considered the possibility that an ele-
mentary molecule might be of polar character, has for many years
regarded nitrogen trichloride as a compound containing positive chlorine,
namely, N---(Cl') •. From the dualistic point of view, it was pre-
dicted that an isomer of this compound might exist, namely, N+ + +(Cl-h.
For many years he has attempted to obtain such isomers, but without
success, and the failure of all similar attempts to obtain the isomers
predicted by the dualistic theory indicates that there is something illusory
in the object of such a quest. .
All of these excursions into the theory of valence seemed to lead
but to an impasse. Thus the dualistic theory, while offering an ade-
quate explanation of the nature of extremely polar substances, and
bringing into relief the electrochemical properties of the elements,

proved incompetent to explain the chemical bond and the behavior of
the relatively non-polar compounds, especially those of organic chem-
istry. Gn the other hand, the pure structural theory gave complete
satis faction in the interpretation of the chief facts of organic chem-
istry. but seemed little qualified to account for those phenomena of a
highly polar type, culminating in the complete division of a molecule
into charged ions. Finally the suggestion of two entirely distinct kinds
of chemical union, one for polar and the other for non-polar com-
pounds, was repugnant to that chemical instinct which leads so ir-
resistibly to the belief that all types of chemical union are essentially
one and the same. Already, however, there were some hints of the
way out of this perplexing quandary. If the properties of substances
could not be explained by the mere assumption of charged atoms,
might they not be explicable if we should no longer regard the atom
as a unit, but rather if we might ascertain where the charge or charges
resided within the atom itself?
FIG. IS.-Two Overlapping Positive Spheres (Thomson).
It was suggested by J. J. Thomson (1907) that two atoms might

be conceived to be held together by electrical forces without electrical
polarization. At that time he regarded the positive part of the atom
as a sphere in which the electrons were imagined to be imbedded, and
he considered the case of two spheres equal in size, overlapping one
another (Figure 15), with electrons situated symmetrically in the re-
gion of overlapping. He says, "In this case there is no difference
in the electrificatlon of the spheres; we cannot say that one is positively,
the other negatively electrified; and if the spheres were separated after
having been together they would each be neutral. , , , We thus see
that it is possible to have forces electrical in their origin binding the
two systems together without a resulting charge on either system."
Thomson made no further use of this idea, and his view of the
positive sphere has been superseded. Nevertheless, the suggestion
that an understanding of chemical affinity must be sought in the
localization of the charges within the atom itself contains the germ
of the final su1:cessful explanation of the chemical bond.
The first 1 to consider the valence electron as attracting simultane-
ously the positive parts of two different" atoms and thus hecoming the
1 The idea of shared electrons was foreshadowed by Ramsay in rgoB.
THE UNION OF ATOMS; THE MODERN DU:1USTIC THEORY 7S
agent which binds atoms together, was Stark. who developed this
idea very extensively in his "Atomdynamik." At tlrst he ('on-
FIG. 16.-Stark's Model of the Carbol1 Monoxide Molecule.
sidered an electron situated within one atom as sending lines of

force to the positive part of that atom and also to the positive part
of another atom. Later he allowed such an electron to move out part
FIG. I7.-Stark's Model of the Ammonia Molecule.
way toward the second atom, and he usually regarded such an electron,
situated somewhere between two atoms, as the equivalent of the chem~
ical bond. However, in two rases, namely in the union of carbon to

carbon and of hydrogen to carbon. he regarded the chemical bond as
due to two electrons whirh jointly tie atom to atom through their
lines of force. Fib'11rCS 16, 17. 18 and 19 show his pictures of the
structure of carbon monoxide, ammonia, the carbon-carLon' bond and
the carbon-hydrogen bond. In the last two illustrations we see the
FIG. lR-Stark's !>..lodcl of the Carbon-Carbon Bond.
first suggestion of an idea which, as we are going to show in the next

chapter, furnishes an extraordinarily simple explanation of the chem-
ical bond, and completely reconciles the divergent views of the
structural and the dualistic theories.
Bohr (1913 II), extending his theory of the atom to systems con-
FIG. 19.-Stark's Model of the Carbon-Hydrogen Bond.
taining marc than one atom, thought of a chemical bond as a ring

of electrons circulating in an orbit with its plane perpendicular to
the line joining the atomic centers. Thus he says, "Such considera-
tions suggest a possible configuration for a water lp-oleculc, consisting
of an oxygen nucleus surrounded by a small ring of four electrons
and two hydrogen nuclei, situated on the axis of the ring at equal
THE UNION OF ATOMS; THE MODERN DUALISTIC 7'HEORI' 7'1
distances apart from the first nudeus, and kept in equilibrium by help
of two rings of greater radius, each containing three electrons. . .."
Each of these rings of three electrons represented a bond. \Vhen,
however, he considered molecular hydrogen and methane, he concluded
that there aTe two electrons in the ring which constitutes the vond, and
thus obtained a picture not unlike that of Stark.
Kossel (1916) employed models of molecules having rings of
electrons common to two atoms. Figure 20 shows his picture of the
nitrogen molecu1e.
FIG. 2o.-Kossd's Model of the Nitrogen Molecule.
Parson (1915) assumed three distinct types of chemical union.

One of these types involved a pair of electrons acting jointly to hold
two atoms together, as is shown Oil the right-hand portion of Figure 21,
which reproduces his picture of the hydrogen molecule. On the other
hand, he assumes a far more complicated process in the union of
two atoms of chlorine, as is shown in the left-hand part of the figure,
where the dots represent valence electrons and the shaded circles rep-
Cl: Hit
FIG. 21.~ParSon's Models of the Chlorine and Hydrogen Molecules.
resent groups of eight (some of these groups of eight stand for what
we now call inner shells).
Thus we see that these authors from very different points of de-
parture arrive at the same conclusion, namely, that two atoms are
held together by groups of electrons held in common by the two
atoms. Moreover each of these authors occasionally considers that
78 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES'
the group of bonding electrons is a group of two, but this was not
regarded by any of them as an essential feature of chemical union.
We are going to see in the next chapter how far-reaching are the
consequences of the simple assumption that the chemical bond is at
all times and in all molecules merely a pair of electrons held jointly
by two atoms.
Chapter VI.
The New Theory of Valence; the Chemical Bond.
At the close of the last chapter we traced the germination of
the concept of electrons held in common by two atoms. It is this
concept which is going to permit us to harmonize those two main
theories of chemical union which formerly seemed so incompatible with
one another. In developing this thesis we must now consider the new
and somewhat revolutionary ideas regarding valence which were ad~
vanced in my paper of 1916 on "The Atom and the Molecule."
We have already noted the preponderating tendency of every atom
toward an arrangement of electrons with eight in the outermost shell.
This arrangement of outer electrons, which Parson and I called the
"group of eight," and which has since been more tersely expressed
by Langmuir as the hoctet," is attained when atoms lose or gain
electrons to form ions.
So when calcium and. chlorine unite, the calcium atom by giving
off two electrops, and each chlorine atom by acquiring one electron,
assume the ionic state in which each atom has the group of eight in its
outermost sheU. However, we have seen that the assumption of such
ionizations as a necessary acC'")mpanirnent to all chemical combinations,
even if it is assumed to be only "intramolecular" ionization, leads to
conclusions which are not reconcilable with the facts of chemistry.
The new theory, 'It'/zich includes tho possibility of complete ionization
as a special case, may be given definite expression as follows: Two
atoms may conform to the rule of eight, or the octet rule, not only
by the transfer of electrons from one atom to another, but also by
sharing one or more pairs of electrons. These electrons which are
held in common by two atoms may be considered to belong to the
outer shells of both atoms.
The Pairing of Electrons.

The discovery that those electrons which are held jointly by two
atoms always occur in pairs led to the realization that the "rule of two"
is even 111()re fundamental than the "rule of eighL" We see at the
beginning of the periodic table that helium with its pair of electrons
has the same qualities of stability that characterize the remaining rare
gases which possess outer octets. Hydrogen may form hydrogen ion
with no electrons, it may form hydride ion by adding one electron
and thus completing the stable pair, or finally two hydrogen atoms
79
may unite to form the hydrogen molecule, in which each atom shares
with the other this stable pair of electrons.
I called particular attention to the remarkable fact that when we
count up the electrons which are comprised in the valence shells of
various types of molecules, we find that of some hundred thousand
known substances all but a handful contain an even number of such
electrons. It is therefore an almost universal rule that the number
of valence electrons in a molecule is a multiple of two.
Certain metallic vapors which are produced at high temperatures
are exceptions to this rule. Other exceptions which are found at
high temperatures are the monatomic forms of hydrogen and the
halogens, while at ordinary temperatures we have nitric oxide. nitrogen
dioxide and chlorine dioxide. with II. 17. and 19 valence electrons.
Such molecules which contain an uneven number of valence electrons,
and which therefore depart from the simple rule of two, I called odd
molecules. Until a few years ago, the above mentloned substances
were the only ones of known molec:ular weight which were proved
to have odd molecules. Gomberg (1900) discovered a type of odd
mole.("ule in triphenylmethyl, and many similar compounds of tri-
valent carbon have since then been obtained. Analogous compounds
of bivalent nitrogen were obtained by \Vieland (1911, 1914), who
also isolated the interesting substante (C6H5)~NO. By analogy I
suggested (1916 II) that corresponding compounds of nnivalent oxy-
gen might be prepared, and investigation has shown that such com-
~unds probably occur, although their existence has not yet been
fully demonstrated.
These odd molecules which form an exception to the rule of two
may be said in the best sense of the old adage to prove that rule, for
they form a class of substances with very singular properties. VV"jth
the exception of nitric oxide, everyone absorbs light in the visible
part of the spectrum, and most of them are intensely colored. In
so far as they have been investigated, they prove to be highly para-
magnetic. They are very reactive and attach themselves to a great
variety of substances. Even a substance so little prone to forming
addition compounds as hexane, forms a compound with triphenylmethyl.
These odd molecules show a great tendency to combine with other
like or unlike odd molecules to form molecules with an even number
of electrons, such as I" lei, (qJ,C) I, (qJ,C) (NqJ,), (qJ,C)" (NO,).
(where qJ represents a phenyl or other aryl group). The resulting
compounds are usually, though not always, colorless. They do not
show the properties indicative of very great "unsaturation" which
characterize the odd molecules.
Under other circnmstances, and especially in a highly polar en-
vironment," instead of a combination of two odd molecules, one loses
and the other gains an electron, thus forming the ions with an even
number of electrons. Thus pure molten iodine conducts the cur-
rent, indicating the existence of such ions as 1+ a"nd 1-. The conductivity
of triphenylmethyl dissolved in liquid sulfur dioxide is presumably
THE NEW THEORY OF VALENCE; THE CHEMICAL BOND 8.
due to the ions tpaC+ and fJ'aC-, Chlorine dioxide dissolves in water
to some extent to give chlorous and chIoTic acid. Similarly nitrogen
dioxide gives nitrous and nitric acids, and even pure liquid nitrogen
tetroxide conducts the current, thus indicating the existence of the
ions NO:/ and N02;-'
If we could isolate other radicals, such for example as the methyl
radical, the peculiarities of odd molecules which we have noted would
become even more pronounced. The fact that the few odd molecules
which have been studied can be isolated at all shows that they pos-
sess to a minimum degree the properties which would be exhibited by
free radicals in general.
The simplest explanation of the predominant occurrence of an
even number of electrons in the valence shells of molecules is that the
electrons arc definitely paired with one another. \Ve have suspected
such a pairing in the inner shells, and even in the nucleus itself, though
that was but a conjecture. The c\'idence of the pairing of electrons in
the valence shell of the typical molecule amounts very nearly to proof.
When the gas NO, polymerizes to form the gas N ,0" it loses altogether
its brilliant color and all the other properties which are characteristic
of the odd molecule, The single molecules show all of the prop-
erties which we ascribe to a loosely bound electron. In forming the
double molecule it seems as though the two odd electrons had been
suddenly clamped together by some mechanism.
Jt is to be supposed that this tendency to form pairs is not a prop-
erty of free electrons, but rather that it is a property of electrons within
the atom, Even within the atom it is not necessary to assume that
electrons always exhibit this phenomenon. For example, in the metals
of the iron group some of the electrons which seem to be in a con-
diction of great mobility show no evidence of pairing. In nearly
all molecules, however, we must consider the electrons as definitely
grouped in pairs. In my first theory of the atom I represented the
normal group of eight electrons by a cube with an electron at each
corner, but the idea that electrons are coupled leads rather to the view
that the stable octet is to be represented rather as a tetrahedron with
a pair of electrons at each corner,
The production of a typical electron pair seems to produce and
to be indicated by a state of stability in which electrons are firmly
bound. We have agreed to consider the electron within the atom as
synonymous with the electron orbit or elementary magnet. The pair-
ing of electrons can therefore be regarded as equivalent to a conjuga-
tion of two such orbits accompanied by the neutralization of their
magnetic fields and the elimination of magnetic moment. We shall
later have opportunity to discuss these questions further,
The Bond.
Two electrons thus coupled together, when lying between two
atomic centers, and held jointly in the shells of the two atoms, I have
B2 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
considered to be the chemical hondo \Ve thus have a concrete picture

of that physical entity, that "hook and eye," which is part of the
creed of the organic chemist.
\Vhen two atoms of hydrogen join to form the diatomic molecule,
each furnishes one electron of the pair which constitutes the bond.
I{cpresenting each valence electron by a dot, we may therefore write
as the graphical formula of hydrogen H: H. So when the atom of
hydrogen with its onc electron unites with the atom of chlorine with
its s(~ven electrons, they produce the molecule represented by H: ~l: .
Two chlorine atoms form the molecule : CI: CI: .
To represent the complete structure ~f th~' chlorine molecule with
its two nuclei and its thirty-four electrons we might draw such a pic-
tUfe as that :;;hown in Figure 22. However, such a two-dimensional
representation cannot adequately show the spatial configuration of
•• ••
•• •• .. ..
•• .~
•• ••
•• •• •• ••
•• ••
FH;. 22.-The Arrangetn('nt of Electron" in the Chlorine :Moleculc. (The
Large Circles I<epresC'llt the Valcllcc Electrons.)
the molecule, nor can we yet locate with any degree of finality the
electrons which compose the atomk kernels. \Ve may be sure, how-
ever, that each of the outer shells should be represented by a pair of
electrons at each corner of a tetrahedron. Such a tetrahedron would
ordinarly be regular only in the case of a symmetrical atom like that
of carbon in methane, or carbon tetrachloride.
H :CI:
H:C:H, :Cl :c:cr:
H :0:
It wiH, however, only rarely be expedient to indicate in the simple
graphical formula any distortion of the octet. although we may assume
that it often occurs. Moreover, we must constantly bear in mind,
as in the case of ordinary lormuke of organic chemistry, that our two-
dimensional representation fails to represent the true stereochemistry
of the molecule, Thus at first sight the formula for water, H: 0: H,
appears symmetrkal, although we really regard the two hydrog';;'" as
THE NEW THEORY OF VALENCE; THE CHq!lCAL BOND 83
not symmetrically placed with respect to oxygen, Qut rather at two

corners of a more or less distorted tetrahedron. This conclusion
has also been drawn from the properties of liquid crystals by Vorlander
22
(19.By).means 0 f thoIS SImp
. Ie assumptton,
. t hat tee
hI · I bond·15 a
lcmlca
pair of electrons held jointly by two atoms, I showed how the various
types of molecules "ranging from the extremely polar to the extremely
nonpolar" could be interpreted. Quoting from my former paper,
"Great as the difference is between the typical polar and nonpolar
substances, we may show how a single molecule may. according to
its environment, pass from the extreme polar to the extreme non-
polar form, not per saltum, but by imperceptible gradations, as 50011
as we admit that an electron may be the common property of two
atomic shells." Whether the phrase "imperceptible gradations" is
strictly accurate we shall discuss later.
The pair of electrons which constitutes the bond may lie between
two atomic centers in such a position that there is no electric polariza-
tion, or it may be shifted toward one or the other atom in order to give
to that atom a negative, and consequently to the other atom a positive
charge. Rut we can no longer speak of any atom as having an
integral number of units of charge, except in the case where one atom
takes exclusive possession of the bonding pair. and forms an ion.
For example we may suppose that the normal state of the hydrogen
molecule is one in which the electron pair is symmetrically placed be-
tween the two atoms. In sodium hydride, on the other hand, we may
regard the bonding pair as lying nearer to the hydrogen than to the
sodium, making the hydrogen negative; while in hydrochloric acid
the bond is shifted toward the chlorine, leaving the hydrogen with a
positive charge. In the presence of a polar solvent the chlorine
assumes full possession of the bonding pair, and we have complete
ionization. I attempted to represent these displacements of electrons
by such formulre as
H:H, Na:H , H :~I:, [H]< + [:~Il
Even a symmetrical molecule like that of H2 or 12 may from time
to time become polarized in one direction or the other as a con-
sequence of the disturbance due to thermal ·motion. When iodine
vapor is heated to a high temperature the molecule breaks in such a
way as to sever the bonding pair, and forms two uncharged iodine
atoms. On the other hand. in liquid iodine a few of the molecules
break ap:;trt in anothe:- manner. The bonding pair remains intact
but remams the exclusIve property of one atom, forming 1+ and 1-,
These two types of dissociation may be represented as follows:
:X:~.:=:(+·!>; :X:X:=:!.+:~.:.
In other molecules some displacement of electrons may occur without
fnll ionization, thus making the molecule more or less polar.
84 VALENCE AND THB STRUCTURE OF ATOMS AND MOLECULES
Bromine, chlorine, fluorine and hydrogen, in the order named, show

a diminishing tendency toward either of the above types of dissociation.
We say that the bond in the iodine molecule is looser than the bond
in the chlorine molecule. We also say that iodine is a more polar sub-
stance than bromine.
The two ideas are not synonymous, but as a rule the molecule is
less polar the tighter the bond. Professor Branch has called my
attention to a certain ambiguity in this regard. When we speak of a
polar substance or a polar molecule we imply either that the molecules
are largely polarized, or that they are readily capable of polarization.
In other words, we imply that the bonding pair is either displaced in
one direction or the other, or that it is easily displaceable, in which
case we may say that the pair is mobile. The two things ordinarily
go together, but this is not invariably so. The molecule of sodium
chloride is highly polarized, but the electron pair is so tightly held
by the chlorine atom as to possess little mobility.
Let me quote again from my paper. "Let us turn now to a prob.
lern in the solution of which the theory which I am presenting shows its
greatest serviceability. The electrochemical theories of Davy and
Berzelius were overshadowed by the 'valence' theory when the attention
of chemists was largely drawn to the nonpolar substances of organlc
chemistry. Of late the electrochemical theories have come once more
into prominence, but there has always been that antagonism between
the two views which invariably results when two rival theories are
mutually exclusive, while both contain certain elements of truth. In-
deed we may now see that with the interpretation which we are now
employing the two theories need not be mutually exclusive, but rather
complement one another, for the 'valence' theory, which is the classical
basis of structural organic chemistry, deals with the fundamental struc-
ture of the molecule, while electrochemical considerations show the
influence of positive and negative groups in minor distortions of the
.fundamental form. Let us consider once for all that by a negative
element or radical we mean one which tends to draw toward itself the
electron pairs which constitute the outer shells of all neighboring atoms,
and that an electropositive group is one that attracts to a less extent, or
repels, these electrons. In the majority of carbon compounds there
is very little of that separation of the charges which gives a com-
pound a polar character, although certain groups, such as hydroxyl,
as well as those containing multiple bonds, not only themselves pos-
sess a decidedly polar character, but increase, according to principles
already discussed, the polar character of all neighboring parts of the
molecule. However, in such molecules as methane and carbon tetra-
chloride, instead of assuming, as in some current theory, that four
electrons have definitely left hydrogen for carbon in 'the first case,
and carbon" for chlorine in the second, we shall consider that in
methane there is a slight movement of the charges toward the carbon
so that the carbon is slightly charged negatively, and that in carbon
tetrachloride they are slightly shifted toward the chlorine, leaving the
THE NEW THEORY OF VALENCE; THE CHEMICAL BOND 85
carbon somewhat positive. We must remember that here also we

are dealing with averages, and that in a few out of many molecules
of methane the hydrogen may be negatively charged and the carbon
positively.
"In a substance like water the electrons are drawn in from hydrogen
to oxygen and we have in the limiting caSe a certain number of
hydrogen atoms which are completely separated as hydrogen ion. The
amount of separation of one of the hydrogen atoms and therefore the
degree of ionization will change very greatly when the other hydrogen
atom is substituted by a positive or negative group. As a familiar
example we may consider acetic acid, in which one hydrogen is re-
placed by chlorine, H,ClCCOOH. The electrons, being drawn toward
the chlorine, permit the pair of electrons joining the methyl and carboxyl
groups to approach nearer to the methyl carbon. This pair of electrons,
exercising therefore a smaller repulsion upon the other electrons of
the hydroxyl oxygen, permit these also to shift in the same direction.
In other words, all the electrons move toward the left, producing a
greater separation of the electrons from the hydrogen of the hydroxyl.
and thus a stronger acid. This simple explanation is applicable to a
vast number of individual cases. It need only be borne in mind that
although the effect of such a displacement of electrons at one end of
a chain proceeds throughout the whole chain, it becomes less marked
the greater the distance, and the more rigid the constraints which hold
the electrons in the intervening atoms."
\Ve have already commented on the interesting substance
C6H:;S020H which sometimes hydrolyzes to give phenol and some-
times to gIve benzene. It is no longer necessary to consider this sub-
stance as a mixture of two tautomeric forms, in one of which the phenyl
group has a unit positive charge, and in the other of which it has a
unit negative charge, for when we write the formula as follows:
:0:
'I':S:O:H
:0 :"
we see that the bonding pair between phenyl and sulfur may be shifted
toward the one or the other, and when the molecule is broken at this
point, the phenyl group, if it retains possession of the bonding pair,
will combine with hydrogen ion, but if it loses possession of the bond-
ing pair will combine with hydroxyl ion.
So also in the case of nitrogen trichloride it is unnecessary to
assume the existence of two distinct forms. Rather we must con-
sider that the bonding pairs between nitrogen and chlorine may occupy
many positions at varying distances from the nitrogen to the chlorine
atoms. All these possibilities are comprised in the formula
:C1 :
:O:N:O:
These remarks wiJI suffice for the present to show how the pair
theory of the chemical bond retains every essential feature of the valence
theory which has proved so valuable in the interpretation of organic
chemistry, while it also interprets the electrochemical properties of
the molecule. The determination of the degree of displacement of
electron pairs toward this or that atom is left as a subject for fur-
ther investigation. We have remarked that the present theory includes
as an extreme case the modern dualistic theory; but while that theory
required the several electrons to be in all cases the exclusive property
of one atom or another, the new theory regards this complete separation
of charges as occlJrrjl1g only in some molecules under some conditions.
Other Features of the New Valence Theory.
In addition to furnishing an explanation of the single bond, and

of the double and triple bonds which will be discussed in the next
chapter, the new theory which I presented makes several other important
changes in our ideas of valence.
It had always been supposed that the atom of oxygen in every
one of its compounds is tied to another atom or atoms by two bonds.
In applying the new idea of the bonding pair of electrons it became
evident that many of the difficulties of the old-fashioned graphical
formula, such as Werner pointed out, were at once dispelled if oxygen
in many of its compounds, and especially in the oxygen acids, were
assigned a single bond. In this way the artificial distinction which
had previously been made between the oxygen acids and the halogen
acids disappeared. This I showed by writing the following formul::e
for the ions of some of the ortho-acids.
'0'
:C?:C.I:~:
]- :0: 1--
[ [ :9:~:?:J
:0
"
: , '0 ..
'.. ,
r '0'
:O:P:O:
] =- :ci:
:~ :~.i:~:
]==
l ":0:"
" ,
[ :0 :
" .
One is struck by the resemblance between these formulre and those
which we must assign to osmiu~ tetroxide, the ion of hydrofluoboric
acid, and carbon tetrachloride, namely,
'·:0 :
:0 :Os:O:
":0: "
rl :F:B'F'
":F:"
:F:
] - :ci ::C:el:
Cl :
":0:"
THE NEW THEORY OF VALENCE; THE CHEMICAL BOND 87
;"":, It was evident from these and similar fonnulce that in a large num-
ber of elements there is a preponderating tendency for the atom to
combine with four other atoms. Thus, giving similar formulre to
ammonium ion and to methane,
H
H:C :H
:ii:
I stated that "when ammonium ion combines with chloride ion the
latter is not attached directly to the nitrogen, but is held simply through
electric forces by the ammonium ion." Indeed it seems probable that
the nitrogen atom is never attached to other atoms by more than four
bonds. The new symmetrical formula for ammonium ion, and the
corresponding formula= for the suhstituted ammonium compounds, are
in complete accord with the stereochemical and other properties of
nitrogen compounds discussed by Werner.
These features of the new theory will be more fully discussed in
later chapters. In my original paper I contented myself with a brief
description of the main results of the theory, intending at a later time
to present in a more detailed manner the various facts of chemistry
which made necessary these radical departures from the older valence
theory. This plan, however, was interrupted by the exigencies of war,
and in the meantime the task was performed, with far greater suc-
cess than I could have achieved, by Dr. Irving Langmuir in a brilliant
series of some twelve articles, and in a large number of lectures given
in this country and abroad. It is largely through these papers and
addresses that the theory has received the wide attention of scientists.
It has been a cause of much satisfaction to me to find that in
the course of this series of applications of the new theory, conducted
with the greatest acumen, Dr. Langmuir has not been obliged to change
the theory which I advanced. Here and there he has been tempted
to regard certain rules or tendencies as more universal in their scope
than I considered them in my paper, or than I now consider them,
but these questions we shall have a later opportunity to discuss. The
theory has been designated in some quarters as the Lewis-Langmuir
theory, which would imply some sort of collaboration. As a matter
of fact Dr. Langmuir's work has been entirely independent, and such
additions as he has made to what was stated or implied in my paper
should be credited to him alone.
Chapter VII.
Double and Triple Bonds.
\Ve have seen that the normal state of a molecule is one in which
each atom of hydrogen has its stable pair of electrons, and each other
atom has its stable group of eight. '''Ie have seen further that in
many molecules this result is attained when two atoms share between
them the pair of electrons which we have called the chemical bond.
There are many cases, however, in which this process of forming
single bunding pairs does not suffice to complete the stable shells
of the seyeral atoms. Thus for example the atom of oxygen has but
six valence electrons, and even if two atoms share an electron pair
this does not suffice for the completlon of two octets. On the other
hand, if two pairs of electrons are shared, then each atom can be said
to have jts group of eight.
\Ve therefore visualize such a union by considering each atom
the center of four electron pairs arranged at the corners of tetrahedra,
the two tetrahedra being joined at the two apices. Such a spatial
arrangement cannot yery well be represented in ordinary type, and for
convenience we employ a more schematic representation of the molecule,
namely, : 0 : : 0: .
This theory is in complete accord with the conventional theory of
the double bond as it has been employed in the development of struc-
tural organic chemistry. Neither in the old nOr in the new theory
is the double bond to be regarded as in any sense the equivalent of
two single bonds. Organic chemists speak of compounds possessing
the double bond a5 unsaturated, and this term implies much mare than
a failure to comply with the simplest of valence rules. Unsaturation
connotes a set of properties, aU of which indicate a looseness of struc-
ture and lack of stability. Unsaturated substances as a class are char-
acterized by a high degree of reactivity, by their tendency to change
into saturated substances through a rearrangement or through the
addition of other substances, and often by the existence of color. The
state of unsaturation in a molecule is often manifested by a tendency
to form loose complexes with other molecules. When we compare
unsaturated with saturated hydrocarbons we note that the former
show a marked tendency to form such complexes; thus a substance
recrystallized from an unsaturated hydrocarbon frequently carries with
it solvent of crystallization. The unsaturated substances are sometimes
said to possess residual affinity.
88
DOUBLE AND TRIPLE BONDS
All of the characteristic properties of unsaturated substances in-

dicate that the electrons have not fallen into those positions of symmetry
and low energy that they seem to occupy in saturated types. One
of the most remarkable properties of ordinary oxygen is its pro-
nounced paramagnetism. This unquestionably is connected in some
very intimate manner with the unsaturated state of the oxygen molecule;
the double bond does not seem to afford that conjugation of electrons
(considered as orbits or magnets) which leads to the self-contained
magnetic system existing in saturated molecules.
Ethylene is the typical unsaturated organic compound, and to it
we may assign the formula ~ ~ which is entirely analogous
H:C::C :H
to the formula we have assigned to diatomic oxygen. Indeed there
is much resemblance between these two substances. \Vhile the sat-
urated compounds of organic chemistry are invariably diamagnetic,
ethylene is distinctly paramagnetic, and the ethylene bond in any more
complicated molecule diminishes the diamagnetism.
Also in their chemical reactions the two substances resemble one
another. It is usually assumed that at ordinary temperatures suh-
stances oxidize first tu form a peroxide. Thus hydrogen set free in
the presence of oxygen forms hydrogen peroxide, and substances of
the type of triphenylmethyl immediately combine with atmospheric
oxygen to produce the corresponding peroxides. Such a process may
be regarded as a breaking of one part of the double bond, and the
phenomenon is entirely analogous to the addition of hydrogen or
bromine to ethylene. In these reactions oxygen and ethylene be-
have almost as though they had structures represented by the formulre,
H H
:0:0: , H:C:C:H,
in which the molecules are not odd molecules, but each atom is an odd
atom, in the sense that it has an unpaired electron. It must not be
supposed, however, that such formulce are anything else than gross ex-
aggerations of the state of affairs which is associated with the double
bond. Perhaps they represent extreme states which may occasionally
be attained by a few molecules.
The properties of unsaturated organic substances have proved
to be in very good accord with the <I strain theory" proposed by Baeyer
(1895). According to this theory, a bond tends to lie in a straight
line joining atom to atom, and the directions of the four bonds emanat-
ing from a single carbon atom are those determined by tetrahedral
symmetry. Whenever the bonds are forced out of these positions there
is supposed to be a state of strain which manifests itself through in-
stability and the general characteristics of unsaturation.
According to this idea it is possible to form a ring of five or six
atoms without producing appreciable strain, but rings of three and
four members require a considerable distortion of the bonds. In fact
ring-compounds such as tetramethylene and trimethylene ~ave many

of the characteristics of unsaturated compounds. In the nng of two
members, in other words in the double bond, the effect reaches a
maximum.
I showed in my paper of 1916 that the precise equivalent of this
idea is obtained in terms of the new theory 'Ii{ we make the simple
assumption that all atomic kernels repel one another, and that molecules
are held together only by the pairs of electrons which are held jointly
by the component atoms. Thus two carbon atoms with a single bond
strive to keep their kernels as far apart as possible, and this condition
is met when the adjoining corners of the two tetrahedra lie in the
line joining the centers of the tetrahedra." Furthermore, if a carbon
kernel is attached to four like kernels, these four by their mutual
repulsion will assume the form of tetrahedral symmetry.
However we choose to visualize this condition of strain, it is evi-
dent that, when two atoms attempt to share not one electron pair but
two, the molecule does not settle into an inert condition of high stability
and low electron mobility, but rather that the system adjusts itself as
it best may under the circumstances, and that in this adjustment either
one or both of the bonding pairs remains in a state in which the electrons
are neither tightly held nor capable of forming with each other a seIf-
contained magnetic system. N or can we conclude that this condition
is confined to the two pairs of electrons which a"re sUPIX>sed to con-
stitute the double hondo The properties of such molecules show very
clearly that the remaining bonds are also affected in similar manner, as
though their electruns were also drawn away from their IX>sitions of
greatest stability and rendered more mobile.
Thus the various phenomena which indicate the breaking of a bond-
such phenomena as the disruption of a molecule or the rearrangement
of radicals-almost invariably occur in the neighborhood of such a
seat of ul1saturation. 1£ not as an accurate scientific analogy, at least
as a metaphor, we may liken the phenomenon to the result of a strain
applied to some portion of a mechanical system of elastic bodies. Other
portions of the system then yield in such a manner as to reduce the
strain at the first point and distribute it more evenly throughout the
system. This process continues until some portion of the system,
of greater rigidity, fails to take part in~ or communicate further. the
general yielding movement. So in an organic molecule a condition
of strain produced in one atom may communicate itself through a
chain of atoms, provided that these all have mobile bonds, but the
effect ordinarily is greatly diminished when an atom of the saturated
type is reached, that is, an atom of rigid bonds. Thus the whole benzene
ring may be cO!lsidered to be in "a state ?f considerable mobility, but
the benzyl radIcal, C.H,CH" behaves lIke a methyl radical.
One of tke most interesting types of unsaturated molecule is one
which contains the 50-called conjngated double bonds of the types
- C = C - C = C -, - C = C - C = O. In such systems the con-
dition of unsaturation seems not to be localized at the two double
DOUBLE AND TRIPLE BONDS 91
bonds, but seems to be shared by that part of the molecule where the
ordina~y fo~ula indicates, the existence of a single bond. This process
of coTIjugatlOn, whatever It may be, appears to diminish the unsatura-
tion and to increase the stability of the molecule as a whole. Thus
we find that when a ~ - y - unsaturated acid is heated with a base
the double bond shifts to the a - ~ - position as follows:
CH = CH - CH, - C - OH -----> CH, - CII = CH - C - OH.
I II I II
R 0 R 0
When such a conjugated molecule undergoes addition. the two
added radicals do not necessarily enter on the two sides of a double
bond, but more frequently the addition occurs in the I - 4 position as
follows:
R - CH = CH - CH = CH - R +
XY _____,
R - CHX - crr= CH - CHY - R.
The explanation of "partial valence" given by Thiele (1il99) can best
be translated into the terms of the present theory by assuming such
a molecule as the following with four unpaired electrons:
H H H H
H:C:C:C :·C:H.
But here again we must regard such a formula as a highly exaggerated
representation, or as a representation of an extreme state of the
moleCUle which only occasionally occurs.
A quite different explanation of the conjugated system may be
obtained by considering the arrangement of the atoms in space. This
theory. which was first advanced hy Erlenmeyer (1<)01), has been once
more brought forward by Huggins (1922), who has developed a
~/
~~
A B c
FIG. 23.-Conjugated Double Bonds (Huggins).
number of interesting ideas of organic structure on the basis of the
new theory of valence. In Figure 23, A represents the ordinary formula
of a conjugated system, while ·B is the picture suggested by Huggins.
The carbon atoms are represented by tetrahedra, the circles showing
the positions of electron pairs.
It is an essential part of Huggins' theory that a pair of electrons
in an unstable condition attracts another similar pair (in such man-
ner, we may suppose, as to diminish mutually their unsaturation, or
to neutralize their magnetic fields). Therefore two electron pairs, one
from each double bond, approach one another until they come to-
gether as in configuration B, in which the distinction between the single
92 V ALENeE AND THE STRUCTURE OF ATOMS AND MOLECULES
and the double bonds has been obliterated. In an addition reaction

the two outer atoms may now he attacked, leaving the double bond
between the two centcr atoms, as shown in configuration C.
In accordance with our present theory, it is to be noted that the
tetrahedra have no significance except in so far as their corners rep-
resent the situation of electron pairs. The positions of the atomic
kernels are not indicated, and it seems to me that if we are to adopt
such a picture of a conjugated system we might well assume that in
such a structure these kernels are pushed out toward the periphery,
50 that the outermost bonding pairs (being brought nearer into line
with the atomic centcrs) possess more nearly the properties of the
single bond. The greater part of the condition of unsaturation is
then localized at the center of the molecule. There also it is possible
that the condition of unsaturatioll may be largely diminished if we
are willing to admit that two or more unstable bonding pairs, when
brought together, tend to neutralize mutuallv their states of unsaturation.
(\Ve are speaking In vague terms, and although we shall attempt
to clarify our ideas somewhat in a later chapter, it seems at present
impossible to give very precise definition to the ideas which we express
in the term unsaturation. But we apparently imply that a pair of
electrons v~"hi("h is unable to reach its normal or stable position is the
renter of a field of fone or "residual affinity" which extends from
the atom. It would perhaps be premature to assume that this field
F1G. :24.-Huggins' Model of the Benzene 1101ecule.
is a magnetic field in the ordinary sense of the term, but it seems to

have the general characteristics of such a field, and to be present
under circumstances in which experiments indicate that the molecule
possesses a magnetic system whose field is not entirely self-contained
or self-neutralized. So it is interesting to note that Pascal (19I2)
finds systems containing conjugated double bonds more diamagnetic
than systems containing the same atoms and the same number of double
bonds, but unconjugated. This seems to show that there is some
process associated with conjugation which diminishes the degree of
unsaturation. )
Such views of molecular structure are especially to be judged by
their success in interpreting the behavior of the symmetrically con-
jugated system of benzene. Here Huggins employs a model of the
type first suggested by Korner in 1874. It is represented in Figure 24.
In this configuration there are three pairs of electrons concentrated at
the center of the ring, and if we assume as before that the outer
bonding pairs possess nearly the properties of single bonds, we have
a picture which goes far toward portraying the benzene molecule

as we know it. \Ve must look to further experiments, especially on
the possibilities of isomerism in substituted benzene, to determine
whether such a configuration is admissible.
It is evident that these models of conjugated systems do not at
aU conform to the Baeyer strain theory (although it is to be noted
that Baeyer himself proposed a benzene structure similar to that
shown in Figure 24). Perhaps it could be assumed that as the
bonds are distorted from their normal position a position of maximum
strain can be reached, beyond which further distortion would produce
a greater stability. \Vhether the assumed force drawing together the
unstable electron pairs is sufficient to cause such distortion seems for
the present to be a matter of pure conjecture.
However, I am inclined to believe that there is some change of
structure of a distinctive character occurring in conjugated systems,
and that it is not unlikely that this change is of the character ad-
vocated by Huggins,l eSlx'cially if we add to his hypothesis the sug-
gestion which r offered above, namely, that in this change the atomic
kernels are displaced in such manner as to give one pair of electrons
in each double bond a character more nearly resembling a single bond-
ing pair. \Vhate\'er models of molecular stmcture we may a~sume,
it must always be borne in mind that these models may at hest rep-
resent the configuration of some molecules some of the time, and that
most molecules, especially those with bonds of the weaker sort, are in
constant tautomeric change from one configuration to another.
The Triple Bond.

In accordance with the practice of organic chemists, we should
represent a triple hond by three pairs of electrons held in common
by two atoms. This is the only way in which we can assign to each
carbon atom in acetylene its full quota of eight electrons. Such join-
ing of two carbon tetrahedra at three apices should produce a very
high degree of unsaturation according to the Baeyer theory of strains.
Indeed it has often Leen assumed that the triple bond implies a
much higher degree of unsaturation than the double bond, but I be-
lieve that this view is erroneous. Bromine adds much more rapidly
to ethylene than it does to acetylene. and Pascal has shown that the
diminution in diamagnetism is far less for the acetylene than for the
ethylene linkage. On the whole it would appear that molecules which
are ordinarily assigned the triple bond show less unsaturation than
those which are assigned the double bond.
Whether, or to what extent, the carbon atoms in acetylene actually
share three electron pairs is an open question. It is to be noted that
1 There is one idea which Huggins employs which seems to me to be unsup-
ported by the evidence. This is the idea that occasionally electrons form stable
groups of three instead of the normal pairs.
while the partial rupture of an ethylene bond in a symmetrical manner

requires the formation of two odd atoms, an acetylene bond may
be partly broken in such a manner as to leave only a single bond,
while all of the electrons remain paired, but both carbon atoms could
no longer retain their complete octets. The properties of acetylene
would best be explained by assuming some intermediate form or some
tautomeric equilibrium between a number of configurations, such as:
H:C:: :C:H and H:C:C·:H.
Limitation of Multiple Bonds.

Some years ago, while compa.ring the properties of dosely related
elements, I was much mystified by the occasional large differences
between substances supposed to he analogous. Thus if we compare
elements of the two periods which end with fluorine and with chlorine,
we see that fluorine and chlorine resemble one another closely in their
physical properties, and the differences that exist are such as might
be predicted from the general trend in the properties of the halogen
group. On the other hand. oxygen and nitrogen differ far morc from
sulfur and phosphorus, and when we compare carbon and silicon and
their comJXlunds we encounter some very curious anomalies.
The tetrachlorides of carbon and silicon are both liquids and closely
resemble each other, although the latter has a higher boiling poin.t, as
might be expected from its greater molecular weight. On the other
hand, two substances could hardly differ more widely than do carbon
dioxide and siJkon dioxide. The former is a gas, the latter is a solid
which does not melt or vaporize except at very high temperatures.
The formulce which have been assigned to these substances are identical,
namely, 0 == C = 0 and 0 == Si :::::: 0, and in no way account for the
great dissimilarity in the physical properties of the two substances.
A consideration of a number of such facts led me some years ago
to conclude that the ability to form multiple bonds is almost entirely,
if not entireiy, confined to elements of the first period of eight, and
especially to carbon, nitrogen and oxygen. If silicon is incapable of
sharing a doubJe bond with oxygen, then in terms of the older valence
theory we should be obliged to represent silicon dioxide as
I
~O-Si-O-
I
and the uncompleted bonds could only be completed by union with
other molecules. In terms of the newer theory this would be indicated
by some such configuration as
. :0 :Si:O:
:0 :si:o:
.. :0 :S;:O:
Such a formula would indicate the existence of unsatisfied valences

which could be satisfied only by a process of polymerization which
might end in the formation of some complex ring structure with all
the valences satisfied, or might lead to an indeterminate molecule, that
is, a single molecule which continues to add to itself until all the ma~
teria! is exhausted. It is probable that silica, both in the crystalline
state and in the state of a supercooled liquid, possesses such a structure,
and that the whole mass may be regarded as one indeterminate molecule.
In terms of the theory of electron pairs we should therefore adopt
'\ some such picture as that shown above, although this particular pic-
ture, having only two-dimensional extension, cannot be supposed to
represent the actual configuration of silica.
If sulfur is unable to form a double bond, the molecule could not
take the form : is: : S:, but rather would tend to form a continuing
molecule in which one of the end atoms would have only three pairs
instead of the normal group oi eight, such as : S: S: S . This deficiency
would, however, be made up, and each atom ·~o~ld··have its octet, if a
ring were formed. In the formation of such molecules as S6 and S8.
and in the formation of polysulfides, we haye strong evidence for the
correctness of this view of the structure of sulfur molecules. So also
the failure to form mUltiple bonds in phosphorus offers an explanation
of the molecule p., and possibly other complex molecules which may
exist in phosphorus vapor.
The indisposition of sulfur to form double bonds is well illustrated
in the organic compounds in which the oxygen of a carbonyl group is
replaced by sulfur. Such substances, in so far as they have been pre~
pared at all, show evidence of great chemical unsaturation. They are
for the most part colored, and show a marked tendency to form asso-
ciated molecules. The peculiar properties of the carboxyl group, in
which the carbon can be replaced by silicon and the oxygen by sulfur,
will be discussed at some length in a later chapter.
My observation that it is the elements with kernels of the helium
type that form double and triple bonds led Eastman ('922) to a very
ingenious theory of multiple bonds. He advanced the idea that the
two electrons of the nucleus, which are ordinarily regarded as occupy-
ing }X)sitions of great stability and taking no part in chemical reactions,
might under certain circumstances act as valence electrons in cases
where they are only one step removed from the valence shell. He
therefore assumed that, in cases where single bonds do not suffice to
furnish the full octets, the octets are completed by the entrance of these
inner electrons into the valence shell.
This is an interesting theory, and not only offers a new picture of
the multiple bond, but affords the only explanation which has so far
been offered to account for the existence of the two hydroborons, B2Ha
and B.H 10• which appear to be so similar to ethane and butane. N ever-
theless, there are some serious objections to the acceptance of East-
man's view, the chief of which is that our X-ray data seem to indicate
that the removal of electrons from the inner shell would require a far
greater expenditure of energy than is available in ordinary chemical
processes. 1t seems not impossible, however, that a modification of
his theory may be useful in which it is assumed that the electrons of
the inner shell remain in this position and, while serving as the inner
shell of one atom, help fill the outer octet of another atom.
Resume
We shall have frequent occasion to discuss the phenomenon of
unsaturation, for it is by no means confined to systems containing
multiple honds. Indeed there are many types of electronic arrange-
ment which lead to degrees of unsaturation which may exceed those
which occur in our multiple bond systems. Probably in any case it is
a mistake to consider the process of forming a mul6ple bond as one
which causes a condition of unsaturatioll. It is more likely that the
reverse is true, and that we should consider the process which forms
the structure known as the multiple bond as diminishing the uJ;lsatura-
tion which would exist if this process, whatever it may be, were in-
hibited.
We shall have occasion again to discuss the structure of these
multiply-bonded systems, although we shall not find it possible to give
any definite and final picture of the electronic coufigurations of these
mobile molecules. For the present it suffices to state that, when there
are not enough electrons in a molecule to provide each atom with its
stable octet by the process of forming normal bonding pairs, two con-
tiguous atoms may, to some extent, share a second or third pair of
electrons, although this sharing is by no means so complete or unam-
biguous as in the single bond; and furthermore that this ability to
share a second or third pair is almost entirely limited to carbon,
nitrogen and oxygen.
Chapt(,f VIII.
Exceptions to the Rule of Eight.
The striking prevalence of molecules in which each atom has its
full quota of four electron pairs in the outermost shell has led Lang-
muir to attempt to make the octet rule ahsolute, and he even proposes
an arithmetical equation to determine, in act'ordance with this rule,
whether a given formula represents a possible chemical substance. I
belieyc that in his enthusiasm for this idea he has been led into error,
and that in calling the new theory the "octet theory" he overemphasizes
what is after all but OTIe feature of the new theory of valence. The
rule of eight, in spite of its great importance, is less fundamental than
the rule of two, which calls attention to the tendency of electrons to
form pairs.
The electron pair, especially when it is held conjointly by two atoms,
and thus constitutes the chemical bond, is the essential element in
chemical structure. Even the rule of two has its exceptions in the odd
molecules, and such molecules containing an uneven number of elec-
trons obviously cannot conform to the rule of eight. But there are
many other exceptions to the octet rule and to my mind these exceptions
instead of weakening the fundamental theory, strengthen it; for the
reactions of such substances are just such as the great tendency to
complete their broken octets would lead us to expect.
The exceptions to the rule of eight may be divided into two classes,
one in which an atom has less than four electron pairs in its valence
shell, and one in which it has a greater number than four. We shall
see that these two types of exceptions are probably very different in
character.
In my earliest speCUlations on this subject in 1902, I thought of
the molecule of sodium chloride as produced by the complete transfer
of an electron from the sodium atom to a chlorine atom, thus giving
to chlorine its valence shell of eight electrons, and leaving sodium only
with its inner shen of eight. The two charged particles were supposed
to be held together by the electric f orees then acting between the
sodium ion as a whole and the chloride ion as a whole. This is the
view which has been recently entertained by Langmuir in order to
maintain the octet rule, which would prevent our ever assigning a
chemical ~ond to. an element like sodium. There is no question that
elements hke sodIUm do have a very great tendency to give up their
electrons and form the ions. When a strong electrolyte is dissolved
97
!)8 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
in water all our evidence seems to indicate that the chemical bond has
disappeared, even though we may assume the existence of some un-ion-
ized molecules. But would this be found to be the case if we should
study more fully the properties of sodium chloride in the vapor state?
In the fused state most typical salts have properties which indicate a
completely polarized condition of the molecule. but fused beryllium
chloride has a very low electrical conductivity and may be regarded as
a weak electrolyte. It is best to assume that the beryllium is tied to
the chlorine by definite though possibly very loose bonds. So sodium
may form weak electrolytes with other anions. It is very likely that
the hydrides of the alkali metals will prove to be weak electrolytes,
and when we consider a substance like sodium methyl we find that it
possesses no saline propetties. Presumably sodium is attached to the
methyl group by a chemical bond and thus possesses an uncompleted
octet.
The compounds of boron are of much interest in this connection.
Our knowledge of these interesting substances has been greatly clarified
by the recent work of Stock (1914, 1921). The alkyl compounds of
boron have vapor densities corresponding to the simple formula BRa.
Thus we may write for boron trimethyl or triethyl the formula
R
R:B
R
There seems to be no way in which boron can have more than three
electron pairs in its valence shelL And the properties correspond to
the above formula, for boron trialkyl is highly reactive toward any
substance which is capable of furnishing the electron pair to complete
its octet. Thus with ammonia it fonus a compound which remains
undissociated when dissolved in benzene. This compound may be rep~
resented by the formula
RH
R:~::r;::H ,
RH
where a free electron pair (or, as Huggins calls it, a lone electron pair) .
of the nitrogen completes the boron octet. In the older valence tbeory
the formation of such a compound would be unintelligible.
Just after tbe publication of my paper of 1916, Mr. A. S. Richard"
son, then a graduate student at Princeton University, called. my atten-
tion to the great interest attaching to the boron compounds in the light
of the new theory of valence, and he undertook an extensive theoretical
and experimental study of the compounds of this element. This work,
however, was interrupted by the war, and unfortunately has not been
resumed. Undoubtedly a fnrther study of the compounds of boron
will furnish information of much value, for the compounds of no
EXCEPTIONS TO TIlE RULE OF EIGIlT 99
other element show so clearly the differences between the old and the
new valence theories.
In the halides of boron we might allow the possibility of completing
the shell of eight about the boron atom, for one of the halogen atoms
might share two pairs of electrons with the boron. Indeed our present
theory does not exclude the existence of doubly bonded halogens. In
iodonium compounds we shall assume that the halogen atom is bonded
to two other atoms, and there seems no reason to believe that fluorine
might not, at least occasionally, oe connected by two bond:; to a single
atom. But in the present instance the properties of the boron halides
are fully exp1ained by assuming a sextet of electrons jn the borol) shell.
Thus water, ammonia, or hydrofluoric acid readily shares one lone
electron pair with the boron to complete the latter's octet. The com-
bination of horon trifluoride with hydrogen fluoride to form hydro-
fluoboric acid is indicated as follows:
:F: :ii:
:ii:B + :ii:H :F:B:F:H
":F: .. :F:"
While molecular weight determinations of the boron trihalides show
that they exist as simple molecules, Loth in the gaseous and in tht: dis-
solved states, It is probable that in the solid state the extra electrons
of one molecule enter the uncompleted octet of another molecule, thus
fanning an indeterminate molecule of the type
:ii: :ii:
:F:B:F:B
":F:" :F:
!"rofessor Hildebrand has called my attention to the probability that

In the case of many solids which readily sublime, in other words which
pass from the solid to the vapor state without the intervention of the
liquid state, there is indication of the formation of such indeterminate
or continuing molecules which exist only in the solid phase.
Entirely analogous to the boron trihalides are the corresponding
aluminum compounds, and aluminum like boron frequently completes
its group of eight by attaching to itself a lone pair belonging to another
atom. Thus the compound between aluminum chloride and ether is
entirely analogous to that between boron trichloride and water. In
such manner we may visualize the various intermediate substances
which presumably are formed in the important reactions of Friedel
and Craft.
There is great similarity between the physical and chemical prop-
erties of boron trifluoride and of sulfur trioxide. Since sulfur does
not readily form a double bond, we must write the formula of sulfur
100 VALENCE AND THe STRUCTURE OF ATOMS AND MOLECULES
trioxide as in the following scheme, which expresses the union of

sulfur trioxide and oxide ion to form suHate ion;
In a similar way sulfur trioxide reacts with ammonia. These are

all typical examples of reactions hetween molecules which have un-
bonded electron pairs, and molecu1es which contain uncompleted octets.
Such processes are very much likt" thl' one which occurs in the formation
of ammonium ion from aml11o-nia and hydrogen ion. Here the nitrogen
furnishes the pc1ir of electrons to complete the stahle group of hydrogen,
which in this case is not a group of eight uut a group of two. Thus
[H]+
-+ H [H :~:H = H:~:HJ
1+
H I-I.
Our picture of the formation of sulfuric acid from the anhydride

and water suggests a possible new type of isomers. The two following
formulc.e A and B may represent respectively the commoner state of
the sulfuric acid molecule, and the state of the molecule when the acid
is first formed.
:0: :0: H
(A) H:O:S:O:H, (l\) :(l:S :O:H
--:0: -- :0:
N ow we shall see numerous cases of this kind in which the opportunities

for isomerism are unfulfilled on account of the great mobility of the
hydrogen nucleus (hydrogen ion), and its rapid exchange from one
electron pair in the molecule to another. This mobility is especial1y
marked when we are dealing with a substance of even the slightest
acid properties. 'Verner has called attention to the possibility of
isomers of the type of A and B, but he was still sufficiently under the
influence of the older valence ideas to call A a valence compound and
B a molecular compound. As we have written the two formulc.e we
see how unjustifiable such a distinction is. In the rapid tautomerism
which we assume to occur between these two forms of sulfuric acid
there is no change in the essential dis-tribution of the electron pairs.
We .merely assume that the two hydrogen nuclei are moving about
from one pair of electrons to another, and are only rarely attached to
the same atom of oxygen. However, we sho~1d expect the mobility
to be very greatly diminished if alkyl groups were substituted for
hydrogen, and by treating an acid anhydride with an ether it is to be
EXCEPTIONS TO TilE RULE OF EIGHT 101
predicted that compounds will be formed which will not immediately

rearrange to give the normal esters. .
Another type of substance in which we may assume a sextet rather
than an octet is represented by metaphosphoric acid. Since we are not
incline'd to assign a double bond to phosphorus, we may use the follow-
ing formulre to represent metaphosphoric acid and its mode of com-
bination with water.
:0: H r:
:0 II]
H:O: P + :O:H H:O:P:O:H
":0: ":0: ..
The resulting m()lecule corresponds to the formula n given for sulfuric
acid. In such forms it is evident that the addition of two hydrogen
ions to the same oxygen should produce a highly polar molecule. If
we assume that the above transition form in the hydration of meta-
phosphoric acid is therefore so unstahle as to exist only in very small
amount, we might thus explain the slowness of hydration of this acid.
Nitric acid and carbonic acid are analogous to meta phosphoric acid.
Here we would be more willing to assume a double hand between
oxygen and carbon or nitrogen. But it has been pointed out by
Latimer and Rodebush (1920) that the crystalline structure of solid
nitrates and carbonates seems to indicate that the three oxygens are
symmetrically placed with respect to the central atom. If this is also
true in the dissolved state we should write, for these ions, formul::e based
upon a sextet of eIectrons, namely,
'0' ] - [ '0' ] - -
[ :O:N:O:
.. .. , :0:(;:0:
.. .. .
Atoms with More Than Four Electron Pairs.
We must next consider a group of compounds in which the central
atom appears to share pairs of electrons with more than four other
atoms. Such substances are PCI" SF, and UF,. At the time of the
appearance of my first publication, when it became evident that there
is no compound of nitrogen in which it is necessary to assume that the
nitrogen atom is attached to other atoms by more than four bonds, it
seemed possible that other similar apparent exceptions to the rule of
eight might be explained away. Dr. E. Q. Adams and I made a full
study of the available facts concerning phosphorus pentachloride in
order to see whether it might possibly be assigned a formula analogous
to that of ammonium chloride, as has since been done by Langmuir.
namely,
:ci:
:Ci:p :ci:
+ 1
[ ":ci" j
We concluded, however, that the chemical facts were entirely opposed

to any such assumption, and that the five chlorine atoms are directly
attached to the phosphorus atom, each by a bonding pair of electrons.
In other words, phosphorus is surrounded by a group of ten electrons.
So also in the hexafluoride of sulfur we must consider sulfur attached
to six bonds and therefore possessing a group of twelve electrons in
its shell.
However, the fact that so few compounds of this type are known,
and these only when the highly negative elements chlorine and fluorine
are present, has suggested to me that the rule of eight may be applied
to these substances in a somewhat restricted form if we amplify our
ideas of the meaning of the valence shell. \Ve have seen the value of
considering a series of energy leve1s leading outwards from the center
of the atom, and we have seen how under some circumstances it is
necessary to assume that electrons pass from one level to another. In
terms of this theory we may suppose that in addition to the level in
which the valence shell normally falls there are outer levels into which
the valence electrons may occasionally be drawn. If we call the normal
level the first valence shell, it seems probably justifiable to state the
universal rule that no atom ever contains more than four pairs of
electron.s in the first valence shell.
If we adopt this hypothesis, we must state that when an atomic
kernel is attached to more than eight electrons, some or all of these
electrons have passed into a secondary yalence shell. Now such a
transfer of electrons to a secondary shell would normal1y require the
expenditure of energy, and therefore such a phenomenon can only be
expected to occur when the entrance of these electrons into shells of
other atoms provides the necessary energy. The energy released in the
completion of the normal valence shell is greater in the more electro-
negative elements-indeed this is presumably what we mean by an
electronegative element~and the maximum energy change in such a
process occurs in the completion of the octet of fluorine. We may
therefore assume that fluorine and occasionally other electronegative
elements are able to draw tbe electron pairs of other atoms from the
primary to the secondary valence shells.
It must, however, be clear that if we eventually find it desirable
to adopt this hypothesis of secondary valence shells, it would be irra-
tional to apply it merely to cases in which the valence electrons of
an atom exceed eight. On the contrary we should probably wish to
apply the same idea to many molecules in which the several atoms
possess eight or even fewer electrons, for example, in molecules with
double honds. However, at present we seem hardly ready for any
systematic use of this theory of secondary valence shells.
It is interesting to consider whether such substances as sulfate ion
or sulfur hexafluoride are little polarized, or whether they are highly
polarized but form a multipole of such symmetry as to be subject to
no strong orienting forces in an electric field. The hypothesis which
has just heen suggested favors the latter view. The properties of
sulfur hexafluoride are very remarkable. It does not behave at all
EXCEPTIONS TO THE RULE OF EIGHT 103
like a typical polar substance. It is a gas at ordinary temperatures,

which shows extraordinarily little chemical reactivity. It neither
hydrolyzes with water nor reacts with molten alkali. Nevertheless, if
we assume that the six pairs of electrons are drawn out from the
primary valence shell of sulfur so that the fluorine atoms have almost
complete possession of the e1ectron pairs, it would leave the sulfur
atom with a large positive charge. Presumably the atoms of fluorine
are arranged about the sulfur atom in positions of much symmetry,
perhaps in an octahedral structure, and these atoms of fluorine, which
partake almost of the saturated character of fluoride ion, may serve to
form a protective layer about the sulfur atom, which guards it from
themical attack.
In the case of phosphorus pentachloride we can hardly assume a
condition of great symmetry, and this substance is far more reactive
than sulfur hexafluoride, and exhibits an entirely different set of pro-
perties. But uranium hexafluoride appears from its physical properties
to be a nonpolar substance. For example, it shows a high volatility
at ordinary temperatures (although uranium tetrachloride does not
volatilize until about 1000° C.), but uranium hexafluoride is highly
reactive chemically, and perhaps this may be explained if we consider
that the very much larger kernel of uranium is less adequately pro-
tected from external attack by the layer of six fluorine atoms.
In addition to the substances which we have been discussing, there
is a large class of complex compounds which have been specially studied
by \Verner, and in which we are led to conclude that an atom is attached
to more than four electron pairs. In such compounds as the cobaltic
complexes where the ion is apparently bound to six other atoms or
radicals by electron pairs, and in the anions such as SiF6-- and PtCl u--,
we may eventually' assume that the bonding pairs lie in the secondary
valence shens of the several central atoms. Such complexes are known
involving as many as eight electron pairs attached to the central atom.
as in K4M 0 (eN) s. Consideration of these complexes, and of the
\Verner coordination number, will he continued in the next chapter.
Chapter IX.
Valence and Coordination Number.
The normal state of an atom is one in which it possesses four pairs
of electrons in the valence shell, these Il.1.irs frequently sen'ing as bonds
of attachment to other atoms. The rule of eight implies that the
maximum number of bonds between a given atom and other atoms is
four. Many types of atoms seem to ha\'c a tendency to employ this
maximum number of honds, or in other words, to use each electron
pair as a bonding pair, but it is possible that this apparent tendency
may be due to the need of electrons in order to furnish to other atoms
their stahle groups of eight (or in the case of hydrogen its stable
group of two).
Thus ammonia is a molecule in which each hydrogen has its normal
group of two, and the nitrogen atom has its group of eight electrons.
This molecule tends strongly to add hydrogen ion and thus complete
its four bonds, but this process might equally well be ascribed to the
tendency of the hydrogen nucleus (hydrogen ion) to assume its normal
group of two. Nevertheless, it seems probable that the stability of
ammonium jon and of other like compounds may jn part be ascribed to
the tetrahedral symmetry of a structure of four bonds attached to a
central atom.
In organic chemistry, where the concept of valence has been most
definite and most useful, the valence of an atom is defined as the
number of bonds which attach it to other atoms, whether these bonds
be single or multiple. There appears to be no longer any reason for
refusing to make this definition universal, so that it may be applied to
inorganic compounds as well. The inorganic chemist's use of the terms
positive and negative valence was shown to he misleading in so far as
it is implied that we can always ascribe to each atom in a compound a
definite JXllar number or stage of oxidation. When it is expedient to
indicate the polar number, this may be done as I have suggested in a
previous chapter. Thus instead of saying that cobalt is bivalent in
cobaltous compounds, we shall say that it is hipositive. When we
speak of its valence we shall refer~ as in all other cases, to the number
of its bonding pairs. Thus free cobaltous ion has zero valence, but
cobaJtous ion,.which is attached to four moJecules of ammonia, is quadri-
valent. In general therefore, we may define the valence of an atom ill:
any molecule as the number of electron pairs which it shares -with
other atoms.
It has been proposed by Langmuir to call this number of bonds the
104
VALENCE ,1ND COORDINATION NUMBER 105
covalence, but he has associated this term with an arithmetical equation

by which he attempts to predict the existence or non-existence of
chemical compounds. Dy fixing the maximum covalence at four he has
been led to regard two compounds like Siet and pas as fundamentally
different in character, and it seems to me that his nIle brings huck
some of the arbitrary features of the older graphical formul:.e w!l.ich
Werner criticized because they so often create artificial distinctions
between entirely analogous substances. We shall therefore employ the
old term valence to express the number of bonding pairs of all atom,
and just as we found it necessary to conclude that some atoms contain
more than eight electrons in their outer shell, so we shall be occasion<llly
required by definition to ascribe to atoms a valence of five or six, or
even eight.
However, to most of the elements of small atomic number it is
suffieient to ascr]he the maximum valence of four. This is the l'hara("~
teristic valence of carbon, and there arc few compounds of this element
in which it exhibits any other valence, although it has usually heen
considered bivalent in carbon monoxide, and it forms trivalent com-
pounds, as typified hy triphenylmethyl. When an atom of carhon is
attached to less than four other atoms we ordinarily assume a sufficient
numher of douhle or triple bonds to give to carhon its full quota l,f
four hands. Usual as this convention is, we may suspect that on some
occasions it is a little arbitrary.
Thus when we consider the ortho-acids (in which the central atom
is surrounded by four oxygen atoms) we might say that in the silicates
the ortho-ion, SiO.j. ----, is the only one which will satisfy the maximum
valence of silicon, since this element is not prone to douhle hond
formation. But in the case of carhonic acid we assume that the ortho
form can lose water without giving up the quadrivaienC'e of carbon,
thus forming ordinary carbonic acid and carbon dioxide to which we
assign the forrnuhe 0 = C(OH)" and 0 = = C 0, Pursuing the
same line of thought with respect to nitrlc and phosphoric acids we
note the existence of the ortho-phosphate ion P0 1- - - , while ortho-
nitric acid is unknown. Here we might say that nitrogen maintains
its quadrivalence through a double bond, giving to nitrate ion the
formula
[ . :~: rJ
:O:N::O:
If, however, the X-ray analysis of crystalline nitrates shows the three
Dxygens of nitrates to be all alike, this argument may he fallacious,
and the corresponding argument regarding carbonic acid would also
Ippear questionable. Whatever the explanation may be, it is evident
that the elements of the first period of eight less frequently reach the
naximum valence of four than the elements of the second period of
eight.
Boron achieves its valence of four at the same time that it obtains
its group of eight electrons in such compounds as HBF. and
106 VALENCE ANn THE STRUCTURE OF ATOMS AND MOLECULES
(CH,),BNH,. It also appears in some mysterious way to att as a

quadrivalent element in the extraordinary compounds B2H6 and B.H10
which Stock believed to be fully analogous to hexane and butane. C,H,
and C,H". This problem we shall discuss again.
Phosphorus, arsenic and antimony, as well as sulfur, selenium and
tellllrium afC quadrivalent in their most highly oxidized acids, and the
same thing is true of bromine, chlorine and iodine in the perbromates,
perchlorates and periodates, although in the latter case there is evidence
of a still higher valence of iodine in salts which seem to be derived
irom H,rO,.
~Th.en the nitrogen atom in ammonia achieves quadrivalence by
adding hydrogen ion to give the molecule
rH.N~.H
l .i{ J
r
we see the prototype, not only of many reactions of nitrogen compounds,
hut also of compounds containing other atoms which possess lone pairs
of electrons. There can be little douht that the four hydrogens of
ammonium ion are symmetrically placed about the nitrogen atom, and
the stereochemistry of ammonium derivatives points definitely to con~
ditions entirely similar to those found in carbon compounds, and indi-
cates an arrangement deviating from perfect tetrahedral symmetry
only so far as may be due to the differences in the individual radicals
which are attached to the nitrogen.
\Vhen phosphine is in the presence of hydrogen ion the phosphorus
attains the valence of four by adding one hydrogen ion to its lone
pair of electrons, forming the symmetrical phosphonium ion PH/.
\Vith the exception of the elements of the argon type, every atom
which possesses one or more lone pairs of electrons shows a tendency
to increase its valence by adding hydrogen ion or some correspond-
ing radical, thus forming a great group of compounds which may
for brevity be called the "onium" compounds. Just as the solubility
of hydrogen phosphide in water is increased by the addition of an acid.
owing to the partial formation of phosphonium ion, so we find that
the solubility of hydrogen sulfide in water is increased. although less
markedly. by the addition of hydriodic acid. While this fact might
be explained without the assumption of chemical combination, the most
probable explanation of the increase in solubility is that hydrogen sul-
fide combines with hydrogen ion forming the first sulfonium ion
according to the reaction
H:.~:H + [Hr = [H:~:Hr

Numerous derivatives of this ion are known, in which organic radicals
take the place of hydrogen. To form the second stilfonium ion (H.S)+"
in which sulfur would have the valence of 4. would require the addi-
VALENCE AND COORDINATION NUMBER 107
!ion of another positive ion to a group already posith'e1y charged.

The formation of such an ion would therefore be oppost.'d by strong
electric forces. It is quite possible that in strong acid solutions the
second sulfonium ion may be produced in small amounts, but there
is no proof of this, nor do we know any of the organic derivatives of
this ion.
Just as ammonium ion is morc stable than phosphonium ion, so we
should expect the ion OHa+, which may be called the hydronium or the
oxonium ion, to be more stable than the corresponding suHoniutn ion.
The fact that the ordinary reactions of chemistry have been studied
largely in aqueous solutions has led us too frequently to ignore the
possible complexes between dissolved substances and water. \"ie have
far more quantitative evidence regarding the ammonia complexes than
we have regarding hydrates. The combination between ammonia and
an aqueous acid is unmistakable, and when hydrochloric acid dissolves
in liquid ammonia we assume that we have a solution of ammonium
chloride. When hydrochloriC' acid dissolves in water, in all prob-
ability a similar though Jooser complex is formed giving the ions
OH,' and C1-. Indeed Latimer and Rodebush (1920) believe that an
aqueous solution of hydrochloric acid would have the properties of a
weak acid were it not for the formation of this hydronium chloride.
This tendency of oxygen to convert its lone electron pairs into
bonds leads to a large number of oxonium compounds, many of which
have been definitely proved to exist, while many others have been
predicated to explain certain reactions. \Vhen ether is added to liquid
hydrochloric acid the electrical conductivity of the latter is greatly in-
creased. The only simple explanation of this fact is that diethyl
hydronium chloride is formed with the ions
[ §:
R:
R]~ ..-
Hand [ : C! : ]
So also ether and bromine give a conducting solution. In this case

we may regard bromine as an e)..'ireme1y weak electrolyte which yieJds
a very small amount of bromous and bromide ions. The bromous ion
having an incompleted octet completes its group of eight by means of
one of the lone pairs of oxygen, just as the hydrogen ion obtains its
stable group of two in the same manner.
This production of a stronger electrolyte from a weak electrolyte
by the formation of complexes is a very general phenomenon, first
pointed out by Abegg and BodIander (1&)9). We now see more
clearly the nature of this phenomenon. Methyl chloride is a very weak
electrolyte, but, when treated with trimethyl amine, the strong tetra-
methyl ammonium chloride is produced, namely t
loS VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
Here the chloride ion is attached to the complex ion by no chemical

hondo The only force holding the two ions together is the electric force
due to their charges, and the substance must be a strong electrolyte.
(When the hydrogens of ammonium ion arc not all replaced by organic
radicals there is the possibility of a chemical bond which we shall
discuss in a later section.)
The normal halide ions have zero valence, but they readily add
hydrogen ion to form the acids, and it is probable that occasionalJy
another hydrogen ion is added to onc of the lone pairs. Thus it is
possible that such small conductance as is possessed by pure liquid
hydrochloric acid may be clue to the formation of what might be called
chloronium chloride, with the formula [H.:.:lCll' [elf. This, how-
ever, is merely a conjecture, and the only compounds of this type
whose existence has been demonstrated are illustrated by diphenyI
iodonium ion,
[ '1': "I,: '1' ]+,

To form the highest iodonium ion such as IR/ + + it would be neces-
sary to overcome very powerful electric: forces, and it is not likely that
such a compound wlll be prepared.
'It is to be noted that the formation of the typical "onium"
ions is a process which differs in no essential respect from other proc-
esses in which hydrogen or other radicals become attached to lone pairs.
Thus, starting with oxide ion. the addition of one hydrogen ion pro-
duces the ion OH-, the addition of two produces water, and the addi-
tion of three produces the hydronium ion. So likewise we may think
of nitride ion being converted, by successive additions of hydrogen
ion, to imide ion. amide ion, ammonia and ammonium ion.
It very often happens that a molecule contains several atoms capable
of attaching hydrogen to form an "onium" compound, and in such cases
there are interesting possibilities of rearrangement or tautomerism as
the h.vurogen is transferred from one atom to another. In the com~
pound of SOs and Hel the hydrogen may remain with the chlorine or
go to the oxygen. So hydroxylamine may be regarded as a mixture
of the two tautomers
H
H:~:?:H, H:N:O:
H H ..
and while the tautomerism is too rapid to permit the isolation of either
of these compounds, when the hydrogen is replaced by the far more
immobile alkyl radicals we find isomers of both types, namely, the alkyl
hydroxylamines and the amine oxides.~
So also phosphorous acid, HsPO s, may be regarded as a mixture of
the two tautomers
VALENCE AND COORDfN.4TION NUMBER 109
H
:0: :0:
H:9:~:9:H, H:O:P:O:H
.. ii ..
Here again organic· derh'atives of both types are known. vVe shall
later have occasion to discutls further this interesting kind of
tautomerism, which indeed is of the same character as the tautomerism
which we have recently discussed in the case of sulfuric acid. The
fact that only two hydrogens of phosphorous acid are readily replaced
by metals, indicates that the second of the above forms is the one
which chiefly exists, and that the hydrogen is bound to the phosphorus
by a much stronger tie than to the oxygen. So also hyp()phosphoTOUS
acid, I-J a P0 2 , has but one replaceable hydrogen. and the other two
hydrogens may be regarded as attached to the phosphorus.
\Verner calls attention to the interesting series of compounds from
phosphate ion to phosphonium ion. These exhibit in a remarkahle
way the quadriva}ence of phosphorus. With minor changes his formulre
are in complete harmony with our system, according to which this
series of compounds may be written as foHows. (The next to the
last compound is not known as such, but gives organic derivatives,
the trialkylphosphine oxides.)
:0: ] - - - [ :Ci:p:Ci:
:Ci:p:CJ:
H ] -- H
:Ci:i;:Ci:
]- r
[ .. :<?:", ": i_):", ii" , l"
H
H:P:O:
ii ..
Bivalent Hydrogen.
It seems to me that the most important addition to my theory uf
valence lies in the sug~estion of what has become known as the
hydrogen bond. The idea was first suggested by Dr. M. L. Huggins~
and was also advanced by Latimer and Rodebush, who showed the
great value of the idea in their paper to which reference has already
been made.
This suggestion is that an atom of hydrogen may at times be at-
tached to two electron pairs of two different atoms, thus acting as
a loose bond between these atoms. Thus it is assumed that two
molecules of water may unite as follows:
H:O:H:Ci:
i-i i-i
IIO VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
The atom of hydrogen attached to the two oxygen atoms represents

the new type of bond. This theory of the hydrogen bond offers an
immediate explanation of a large number of complexes involving water.
ammonia and other such compounds, which have hitherto been in-
explicable by any kind of stmctural formula.
For example, we know that hydrofluoric acid consists very largely
of double molecules. In its acid salts the ion is HF2 -, but in its neutral
salts the ion contains but onc atom of fluorine, F-. We know nothing
which would lead us to suspect that two normal fluorine ions, each
of which has its completed octct, would combine with one another.
The whole phenomenon, however, is very simply explained if we
admit the possihility that an atom of hydrogen ties together the two
fluorine atoms, as indicated in the following formulre:
It is a striking fact that the liquids whose behavior indicates a large

measure of association, and which have the high dielectric constant
and ionizing power which seem to be an accompaniment of this as-
sociation, are compounds which contain hydrogen as well as lone e1ec~
tron pairs. Such substances are water, ammonia, hydrogen peroxide.
hydrocyanic acid and hydrofluoric acid. The postulate of the hydrogen
bond certainly offers a very simple picture of the association of such
molecules.
Ammonium hydroxide is 8. weak electrolyte, but tetramethyl am-
monium hydroxide is a strong electrolyte. It has usually been sup-
posed that this difference is due to the fact that ammonium hydroxide
is largely dissociated according to the reaction NH4.0H = NH3 + H 2 0,
a type of dissociation which cannot occur in the case of tetramethyl
ammonium hydroxide. It was believed that the NH 4 0H existing as
such would have all the properties of a strong base. Latimer and
Rodebush offer an alternative explanation. They write the formula
for ammonium hydroxide as
H
H:N:H:O:H
H
In this picture the nitrogen is still quadrivalent, but the molecule as a
whole is held together not merely by the electric force between two
oppositely charged ions but by definite characteristic bonds, and the
weakness of the base is due to the difficulty of breaking the hydrogen
bond. On the other hand it is not assumed that the hydrogen of the
methyl group is capable of acting as a bond, and therefore tetramethyl
ammonium ion and hydroxide ion are held together only by virtue of
their opposite charges.
In the preceding chapter I have suggested that when an atom is
attached to more than four electron pairs, these are not in the first
valence shell but in a secondary (or coordination) shell. The same
VALENCE AND COORDINATION NUMBER III
thing may be true of hydrogen (in which the stahle shell is composed
of two electrons instead of the normal group of eight). Hydrogen
when attached firmly to a pair of electrons, as in the hydrogen-hydrogen
or hydrogen-carbon bond. shows no tendency whatsoever to become
bivalent. or in other words, to form a hydrogen hond. On the other
hand, when comhined with an extremely ne~ti\'e element like nitrogen,
oxygen or fluorine, toward whi{'h the electron pairs arc very tightly
drawn, the bonding pair may be drawn out of the first valence shell of
hydrogen into a secondary shell, and it is apparently under some such
circumstances that the hydrogen atom can form a loose attachment to
another pair of electrons, thus forming the hydrogen bond.
The Quadrivalence of Nitcogen.

In the older theory of valence nitrogen atoms with five bonds were
a part of the regular stock in trade of the chemist. When I claimed
that nitrogen is never more than quadrivalent I did not mean to deny
the theoretical possibility of a compound such as NF~, in which the
nitrogen atom would be bonded to five other atoms, corresponding to
the compound PCl, which we have already discussed. What I main-
tained, and still maintain, is that in any compound now known the
nitrogen atom 15 never attached to more than four bonds. \lVe have
seen that in terms of the new valence theory the double bonds which
were formerly used between nitrogen and oxygen, and which required
the quinquivalence of nitrogen, are now for the most part replaced by
single bonds. So also we have seen that in salts of the ammonium
type, like tetramethyl ammonium chloride, the anion is not attached
to the cation at all, and that if it is attached to the cation in the simple
ammonium saits, it has a bond to a hydrogen but nnt to the nitrogen
atom.
The amine oxides form a very interesting class of nitrogen com-
pounds. The old and the new theories are represented by the formuire
R,N = O. and R,N - O. Such oxides when treated with an alkyl
iodide form iodides which are strong electrolytes. It was shown by
Meisenheimer (1913) that. when the iodide ion 15 repia1:.ed hy an
alcoholate ion, isomeric substances can be prepared, to which he as-
signed formulre which afe in complete accordance with our present
views, namely,
[R,NOR]+ [ORT and [R,NOR']+ [ORr.
On the other hand, L. W. Jones (1914). using the modern dualistic
theory, and the idea of quinquivalent nitrogen, wrote the formulre with
both oxygens attached directly to the nitrogen. and called the two sub-
stances electromers; for he believed that one oxygen was negatively
charged and the other positively charged, and that one or the other
isomer is produced according as the positive atom of oxygen is attached
to R or to R'. I understand, however. from Professor Jones that his
prevent view of the structure of these isomers would essentially coincide
with my own.
lIZ VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
At the same time I am obliged to admit that I made privately a

prediction regarding these compounds which I am now forced to retract.
Regarding the positive ion R;!NOR+ as analogous to tetramethyl ammo-
nium ion, and unwilling to admit the possihility of an anion attached by
a fifth hond to nitrogen, I predicted that both the salts and the hydroxide
of this cation would prove to be stron~ electrolytes. In work by Pro-
fessor T. D. Stewart and Dr. Sherwin Maeser which has not as yet
been published, this question was investigated, and it was found that
compounds of the type (R,NOR)I behave in aqueous solution as
typical strong electrolytes, but that they hydrolyze, showing that the
corresponding base is weak. It was also found that in ahsolute alcohol
th('se salts absorh hydrogen jon when an acjd js added. These observa-
tions which at first sight seem haffling are very simply explained on the
theory of the hydrogen hondo
Let us for the sake of simplicity revert to the aqueous solutions
of amine oxide itself. These oxides combine with two molecules of
water to form very stable compounds,-a fact which obviously is not
explained by assuming nitrogen to be quinquivalent. Let us assume
that these two molecules of water form an "onium" compound with
the oxygen rather than with the nitrogen. vVe may then picture the
amine oxide dihydrate as follows:
:O:H
R H
R:N:O:
R H
:O:H.
Such a compound, in which each hydroxyl radical is held to the
remainder of the molecule by a definite bond, would be expected to
act as a weak base. This is in accordance with the facts. In other
words, the weakness of the base is not caused by the attachment of
one of the hydroxyl groups to a fifth bond of nitrogen, but rather by
attachment to oxygen through the hydrogen bond. \tVhen the amine
oxide is dissolved in absolute alcohol instead of in water the formation
of the oxonium complex may be presumed to be far more limited,
and when acid is introduced hydrogen ion may be added directly to a
r
lone pair of the oxygen, forming the ion.
[R:~:g:H
Another group of substances which is particularly interesting to
the theory of quadrivalent nitrogen has been obtained by Schlenck and
Holtz (1917). The first of these substanceS has five hydrocarbon
radicals to a single nitrogen atom, namely, four methyl groups and
one triphenyJmethyl group. This substance, however, was found to

be an electrolyte. and is doubtless to be regarded as a salt with a
triphenyhnethyl anion. namely, [N (CH, ),]' [C(C,H,h]-. They
obtained similar compounds, in which diaryl amino groups replace the
triphenylrnethyl. which also gave conducting solutions in pyridine.
They concluded that the combination l'is of a salt-like or ionogen
character," They further remark, "We thus see a confirmation of
the assumption that in the ammonium compounds the fi fth valence
of nitrogen is under aU circumstances otherwise constituted than the
four other affinities." Finally they obtained a compound which ac-
cording to its method of preparation and its analysis has the formula
(C,H,CH,)N(CH,) •. This substance may be the nearest approach yet
obtained to a comJX>und with quinquivalent nitrogen, but no solvent
could be found in which the substance dissolved without decomposi-
tion, and it therefore could not be very fully investigated.
In stating that no quinquivalent compounds of carbon or nitrogen
are known, we should not exclude the possibility that unstable sub-
stances of this type may appear in a transitory form during the progress
of a chemical reaction, and indeed there are cases in which the
mechanism of a reaction may best be interpreted hy the assumption
of a fleeting addition compound to an atom of carbon or nitrogen that
is already quadrivalent. This is especially true when two of the
bonds constitute a douhle bond, hut may also be assumed when the
central atom is attached to four other atoms. As an example let us
consider very briefly the very interesting phenomenon known as the
'Vaiden in1·crsion.
This phenomenon is observed when a dextro-rotary substance is
passed through a cycle of processes resulting finally in the lrevo-rotary
form of the original substance. There seems to be but one possible
way of accounting for this peculiar behavior. Let us consider a
carbon attached to the four radicals R 1 , R 2 , R lI • and R 4 , and let us
assume that a fifth group R, becomes temporarily attached to the
carbon atom near to the face of the tetrahedron which is opposite to
R,. A slight shift of the kernel might make it now the center of a
new tetrahedron with corners at R21 R". R.tl and Rill while Rl would
become detached from the molecule. Then if the radical R, in the
new molecule were to be replaced by the radical R" the resulting
mo1ecule wou1d be the mirror image of the one with which we started.
In this explanation it is not necessary to assume that the five radicals
are attached to the carbon for any appreciable period of time, indeed
it might be assumed that the R, leaves at the same instant that the
R, becomes attached to the carbon atom.
Valences Higher Than Four.
In substances like phosphorus pentachloride and sulfur hexafluoride

we find a central atom attached to more than four other atoms. We
have mentioned a similar case in per-iodic acid. Indeed iodine fre-
1I4 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
quently shows evidence of holding six pairs of electrons instead of

the normal four. Thus we ha \'e the group of compounds which con-
tain what is ordinarily and I think correctly called trivalent iodine,
of which the simplest example is hydrogen tri-iodide. Here we may
assume that one atom of iodine is the central atom, is attached to
the hydrogen and to the two other iodine atoms by bonds, and pos~
sesses altogether six electron pairs. In the same way we may in~
terpret iodine trichloride and numerous other inorganic: and organic
compounds of iodine.
The majority of cases of valence higher than four are found in
the complexes formed by metallic salts. One of the greatest services
rendered to chemistry by Vverner consisted in the classification and
elucidation of such complexes. He showed that an atom such as the
atom of chromium or cobalt lies in the center of a coordination zone
in which a certain number of radicals are attached to the central
atom, and this number he called the coordinat;on number.
All of the radicals which enter into such coordination groups are
substances which possess lone pairs of electrons, like H 2 0, NH 3 , N0 2-,
CI- and the like, and doubtless one of these lone pairs is employed
in attaching each radical to the central atom. Thus free cobaltic
ion, Co+ + +, has no valence electrons, and when it combines with six
molecules of ammonia to form an ion of the Hhexammin" type. we
may consider the cobalt atom as attached to six electron pairs, prob-
ably situated at the corners of a regular octahedron. Each molecule
of ammonia thus furnishes one of the bonding pairs, and by our
broad definition of valence we may say that the cobalt atom has the
yalence of six. In other words. valence and coordination number are
the same thing. So also, if we choose, we may say that the coordina-
tion number is four for sulfur in the sulfates, is four for carbon in
most 0f its compounds, and is four for nitrogen in ammonium salts.
\Vhether we speak of the valence or the coordination number,
there are two ideas that must be distinguished. It is sometimes cus-
tomary to speak of the valence or the coordination number of an
element without referring to any particular compound. The normal
or maximum valence or coordination number of the element is then
meant. \\Then we are referring to a definite molecule the valence or
the coordination number expresses the actual number of bonding pairs
attached to the atom in question. Thus it is occasionally convenient
to speak of the coordination number of carbon as four or of the
cobaltk ion as six, although there are some compounds in which this
maximum number of bonds is not employed.
Many of the varied phenomena shown by the complexes formed
with metal salts can be ascribed to the rep1acement of one radical
by another, and if one of the radicals is a neutral compound and the
other is an ion, an interesting series of compounds results which
may be illustrated by Figure 25, which is taken from Werner, and
which shows the molecular conductance, at a' concentration of 0.001
normal, of a series of complex platinous salts. Here the coordination
number is four, and, as the ammonia molecules are replaced by chloride
ion, the complex changes from a doubly charged positive ion to a

doubly charged negative ion. The substance in the middle of the
series is uncharged and is essentially a non-electrolyte, as shown in
the diagram.
Often such complexes are held together very firmly, and the process
of replacement of one atom by another is a very slow one. Never·
theless, the general character of such compounds seems to support the
idea that the bonds operating in these complexes ar(: not identical with
our typical bonds, such as the one joining carhon to carbon. In other
words, we may again assume that the electron pairs do not lie in the
primary valence shell of the central atom, but in a secondary valence
shell.
FIG. 25.-Conductivity of Complex Platinous Salts (Werner).
If this idea proves to be correct, we might make a distinction be-

tween valence and coordination number. We could can a primary bond
a valence bond, and a secondary bond-one in which the electron
pair is not in the primary valence shell of each of the atoms which it
unites-a coordination bond. But if this were to be done it would be
absurd to consider that coordination bonds exist only in cases where
more than four bonds are attached to a single atom. Rather it would
be necessary to class as coordination bonds many that are now con-
sidered as ordinary valence bonds. I believe, however, that our present
knowledge is insufficient to permit any systematic discrimination of
this kind.
It will therefore suffice to point out the prohability that the bond-
ing electrons, which hold the several radicals to a central atom of a
metallic element in the coordination zone, are not ordinarily in the
valence shell of that atom. Nor do metals as a rule seem able to hold
electrons tightly. There are, however, some exceptions, and mercuric
mercury, in its salts and in its various compounds with organic radicals,
seems to form a bond which is not unlike the typical bonds of organic
compounds.
u6 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
Valenee in Condensed Systerns.

Hitherto we have been considering the molecular structure of only
those substances in which the molecule is a well defined entity, or in
other words, which possess a definite molecular weight. This win also
be our practice in succeeding chapters, and we shall not ordinarily
base our valence rules upon information concerning the stoichiometric
relationships in solid substances. For example, it would be undesirable
to consider that one of the atoms in cadmium sulfate has a coordina-
tion number of 8/3 merely because the hydrate has the empirical
composition CdSO".8/3H20. Rather we must assume that in this
crystal there is an arrangement of the particles in which 8 molecules
of water enter to every 3 atoms of cadmium.
Nevertheless the structure of crystals and other condensed systems
must eventually come within the scope of valence theory, and we
shall devote the remainder of this chapter to a brief consideration
of such problems.
In liquids, especially of the "sticky" kind, and in glasses, the simple
molecules which exist in the gaseous state presumably become attached
to one another and often form chains or groups in which the in-
dividuality of the simple molecule is partly or wholly lost. Such groups,
or continuing molecules, probably have an extent which is determined
by adventitjous circumstances. On the other hand in crysta11ine sub-
stances these continuing molecules may be conceived as coterminous
with the complete crystal, and characterized by a well-ordered arrange-
ment, except where the crystal is distorted or broken.
Through the study of crystalline form, and especially through re-
cent investigations concerning the reflection of X-rays, we have ob-
tained important information regarding the structure of crystals. We
are able to fix the positions of the heavjer atoms with much certainty;
the lighter atoms are located with greater difficulty, and hitherto we
have been unable, except by inference, to ascertain the positions of
the electron pairs which constitute the atomic valence shells.
However, we already know enough regarding crystal structure to
state that there is in general a great difference in the laws of chemical
combination and in the arrangement of electrons and atomic kernels.
between crystaJljne substances on the one hand and gaseous substances
on the other hand. In the latter case those forces which we know
as chemical affinity must be wholly or largely satisfied within the
confines of a limited molecule; in the former these forces may mani-
fest themselves as so-called crystalline forces operating through a con-
tinuing ttlOlecuJe of unlimited size.
In our discussion of silicon dioxide in Chapter VII, we have shown
how such a oontinuing molecule might arise merely through the forma-
tion of ordinary chemical bonds, and there are doubtless a number of
crystals which may be adequately treated in this manner. The most
striking case is furnished by diamond, in which each carbon atom is
equidistant from four similar atoms, and there can be little doubt
that it is joined to each by a bonding pair of electrons. Diamond

may therefore he regarded as a satorated organic compound in which
each carbon atom is joined by a typical bond to four other atoms,
and it is not unlikely that the rigidity of the bond and the distance apart
of two bonded carbon atoms are about the same as in the molecule of
ethane.
As a rule, however, crystalline structures are not governed by
ordinary valence rules. This problem has been discussed in a very
interesting manner by Langmuir (1916). In a crystal of sodium
chloride each sodium atom is positively charged and is equidistant
from six negatively charged chlorine atoms; and also each chloride ion
is surrounded by six sodium ions. If the molecule of sodium chloride
in the gaseous state condenses upon a crystal of sodium chloride the
gaseous molecule completely loses its identity and the atom of sodium
belongs as much to five other chlorine atoms as to the one to which it
was previously joined.
Another class of crystals is composed of molecules which are
chiefly saturated internally, so that each molecule becomes the unit of
a crystal, in which the several molecules are held together by only a
slight residuum of chemical affinity. Solid argon, nitrogen and hexane
may be taken as examples of this class. Here the same molecule that
exists in the gaseous state preserves its identity in the crystal. The
same two atoms of nitrogen which condense together when the gas
is solidified will reappear together when the solid is vaporized. An
optically active organic substance is not raC'emized by alternate crystal-
lization and melting.
Presumably a substance like carbon tetrachloride would preserve its
molecular identity on crystalliza60n, but whether this would also be
true of carbon tetraiodide seems less certain. There is much inter-
esting work to be done in testing experimentally this problem of the
preservation of the identity of molecules in crystals. For example,
it would be very interesting to know whether solid iodine keeps to-
gether the two atoms which constitute the gaseous molecule.
There obviously must be many gradations between the two ex-
treme types, namely, the crystal in which the primary chemical forces
are largely satisfied inside the unit molecule, and the one in which
the chemical bond, or something very like it, is operating as a crystal-
line force. On the one hand we have very soft crystals like paraffin
or carbon tetrachloride, on the other extremely hard substances
like diamond and quartz. Intermediate are saline substances in which
there are, practically speaking, no chemical bonds, but in which the
atoms are strongly held together by the electrostatic force between
oppositely charged particles. The difference between the extreme types
is shown also in a comparison of melting and subliming points. A
small degree of therm.,1 agitation is sufficient to destroy a crystal of
paraffin or carbon tetrachloride, but a very high tempera tore is needed
to melt or sublime substances like quartz and diamond.
A substance of the latter type, in which the chemical affinities are
largely satisfied, not by the formation of small units, but by the
,,8 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
packing together of layer upon layer of atoms, must exhibit at the

crystal surface a very high degree of chemical unsaturation, and it
is upon such surfaces that we must look for powerful and specific
adsorption of molecules which are capable of partially relieving that
condition of unsaturation. The great amount of heat which is set
free when water is added to finely divided silica illustrates the magnitude
of the energy changes accompanying such a process.
Resume
Defining valence as the number of bonds which are attached to a
given atom, or the number of electron pairs which that atom shares
with others, we may regard the maximum valence of four as a sort
of norm. Howe\'er. we find a great many cases in which an atom
shares mOfe than four pairs with other atoms and, although we sus-
pect that in aH such cases the electrons are drawn farther from the
central atom than in the typical bond, we are not yet able to make a
definite classification on this basis. For the present we must regard
valence and coordination number as synonymous. Hence there are
many elements to which mllst ascribe a higher valence than four and
the number will doubtless be increased by further investigation. Thus
Professor \V. A. Noyes has called my attention to the fact that the
dihydrate of perchloric acid is more stable in the vapor state than the
anhydrous acid; this indicates very strongly a coordination compound
in full analogy with pedodic acid. Passin}y also the dihydrate of
suliurjc acid is not an ordinary hydrate formed through hydrogen
bonds but rather has six hydroxyl groups attached to sexivalent sulfur.
Chapter x.
Compounds of Elements with Small Kernels.
Although the kernel of the atom usually takes no direct part in
chemical combination, its size and structure must play a large part
in determining the behavior of the valence electrons. The degree of
cumplexity of the kernel determines the distance of separation of the
several valence pairs, and the forces which evidently oppose the close
approach of two valence shells Of two kernels to one another, make
it possible to speak roughly of the size of the atom.
Indeed in a crystal at low temperatures the molecules, the atoms,
and the constituent parts of the atoms, may be regarded as occupying
definite positions, and the structure of the unit crystal may be considered
as repeated with complete regularity throughout the whole crystalline
mass. From the study of the reflection of X-rays it has already be-
come possible to ascertain the location of the several nuclei and to draw
some inferences as to the location uf the several electrons. Further
refinements of this method will doubtless permit the complete localiza-
tion uf the valence electrons, and lXJssibly the arrangement of the
kernel electrons as well.
By this method, and through the aid of the known densities of
substances, Bragg (1920) has made some calculations of the radii
of the outer shells of various atoms. Some of the values which he
thus obtained are given in the accompanying table, in which are com-
pared the radii obtained by such a study of the several substances in
their crystalline state with the radii of the corresponding molecules in
the -gaseous state, calculated from the gaseous viscosity by Rankine
(1921 ).
Both methods of calculation rest on some assumptions, but the values
doubtless show the approximate dimensions of the several molecules
and the relative size of the light and heavy molecules. We see that
the dimensions of the atom increase as the number of electron shells
is increased, but the effect is partly offset by the gradual drawing in
of the several shells as the positive charge of the nucleus (the atomic
number) increases. so that the radius of xenon is less than twice as
great as the radius of helium.
A far more accurate method of determining molecular radii is
promised in the present intensive study of the ultra red spectrum of
gases. I cannot enter here upon a discussion of the method of thought
employed in these investigations. and shall only mention one or two
of the data obtained. I am indebted to Mr. H. C. Urey for a list
119
"" VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
TABLE.
MOLECULAR. RADII IN THE CRYSTALLINE AND IN THE GASEOUS STATES.
Radius Radius Radius Radius

(1O-8 cm ) (Io-"cm) (10-8 em) (IO-Scm)
Bragg Rankine Bragg Rankine
------------,II=-----I---;--il--
Helium 0·94 HF 0.65 1.17
Neon .,. ~.65 1.17 HCl 1.02 1.43
Argon 1.02 1.43 HB, " I.I7 1.58
Krypton'" . 1.17 1.59 HI ....... 1.35 1·75
Xenon 1.35 1.75
COMPOUNDS OF ELEMENTS WITH SMALL KERNELS .2.
of the latest results obtained by himself and others through this method.
From the work of Kratzer ('920) one-half the distance between the
atoms is 0-46 x 10- 8 for HF. 0.63 x 10-< for Hel, and 0.70 x 10-' for
HBr. Some other molecular dimensions obtained by this method will
be mentioned later. In the meantime it will be noted that the values
are considerably lower than those given in the table, although they
are arranged in the same order with respect to the several substances.
A very recent paper by Bragg (1922), in which he discusses some
results obtained by X-ray reflection, casts some doubt upon the ar-
rangement of electrons in shells with eight in the outer shell, as pro-
posed in my paper of 1916. His work leads him to believe that a
smaller number of electrons than eight exist in the outer shells. I
think, however, that his observations would have been otherwise
interpreted if he had taken into consideration the fact that the atoms
in a crystal at ordinary temperatures are in a state of motion. with
amplitude sufficient to make the volume considerably greater than it
would be at the absolute zero, It would be desirable if his experiments
could be repeated with crystals at very low temperatures.
Hydrogen and Helium.
The kernel of the hydrogen atom is unique in that it is com-
posed merely of the nucleus, and therefore can be considered to have
practicaUy no vulume, and this singularity is manifested in the chemical
behavior of hydrogen. Unlike all other atoms which enter into chemica1
compounds, its stable shell is nut a group of eight but a group of
two. When it acquires exclusive possession of two electrons it be-
comes the hydride ion, H-, which has the same structure as the neutral
atom of helium and the positive ion of lithium. Hydride ion, as we
have remarked in an earlier chapter, is to be regarded as entirely
analogous to the halide ions, in each of which the stable group has
been produced by the addition of one electron.
When hydrogen is attached to a bonding pair it possesses more
the character of a negative than of a positive element. Thus without
committing ourselves to any quantitative statement, we may safely
say that methane is more like methyl chloride than it is like sodium
methyl.
The hydrogen nucleus may be considered as lying very close to a
bonding pair, and if we examine the figures of the above table we
see that hydrofluoric acid which, except for the hydrogen nudeus,
has the same structure as neon, is also given the same radius by both
of the experimental methods. The same thing is true of the other
acids and the corresponding rare gases. The exactness of the agree-
ment is perhaps accidental, but it seems likely that a molecule of
hydrogen halide has about the same dimensions as the halide ion
alone would have.
When hydrogen loses its electron pair and becomes hydrogen ion
it is altogether without electron shells and, as we have remarked, it
is practicaUy without volume. It therefore does not suffer from those
"sterk" effects to which aU other atoms are subject, and may there~
fore be presumed to be capable of an exceptional degree of mobility.
Indeed many types of tautomerism or rapid rearrangement, both in
inorganic and in organic chemistry, may be ascribed to the movement
of the hydrogen nucleus, and we shall see that in many cases the
essential structure of a molecule can best be understood if we leave
entirely out of consideration those atoms of hydrogen which are easily
removed as hydrogen ion.
When an acid dissolves in our ordinary solvents it is to be sup-
posed that hydrogen ion does not exist largely in the free state, but
rather is held to the solvent in an "onium" complex, Nevertheless,
we are tempted to explain some of the peculiar properties of acid solu-
tions by assuming that the hydrogen ion is occasionally in the free state.
The atom of helium, which is also characterized by the stable
group of two, is also electrically neutral when in exclusive possession
of such a group, and chemists have found no evidence of its chemical
combination with other elements. On the other hand, the band spectrum
of helium is interpreted by the physicists as due to a molecule con~
taining two helium atoms.
It was lirst observed by Rydberg that the spectrum of helium ap-
pears to emanate from two different substances, which were caned
helium and parhelium. It is now known that the different spectral
series are due to a single element, presumably in different states, and
it is assumed that under the excitation of electric discharge an atom
of helium may assume a form in which one electron is in the first
sheH and one is dislodged to an outer shell. In such case the outer
electron would be expected to behave as a yalence electron. The atom
would thus resemble the hydrogen atom and might form compounds,
not only with other similar atoms, but also with a halogen or a metal.
It would be interesting to see whether by exciting helium in the pres-
ence of sodium or iodine, maintained at low temperatures, it would
be possible to isolate these extremely unstable compounds.
Lithium and Beryllium.

The elements of the first period of eight are characterized by a
kernel which contains a single pair of electrons. Structures built up
of such kernels could hardly be expected to exhibit properties which
are altogether similar to those of structures made up of kernels, pos-
sessing a complicated three-dimensional structure. In fact the
properties of the elements in this lirst period of eight are not always
predictable from those of similar elements of higher atomic number.
Thus, to give a single example) the standard electrode potential, against
the ion in aqueous solution, is greater for ruhidium than for potassium,
greater for potassium than for sodium, but the value for lithium is
higher even than that for rubidinm ..
The great tendency of lithium and beryllium ions to form hydrates
distinguishes them in degree though not in kind from other similar
COMPOUNDS OF ELEMENTS WITH SMAl.L KERNELS 12J
elements. Whether they form definite coordination complexes in the

Werner sense has not as yet heen ascertained. l This could be ascer-
tained by a more thorough study of solutions in non-aqueous !'.olvents.
Thus through a study of the solubility of lithium salts ill liquid am-
monia in the presence and absence of water. or of the potential of
lithium amalgam against a solution of a lithium salt in liquid ammonia.
with and without added water, we might obtain the desired information.
The salts of these metals in concentrated aqueous solution show
signs of hydrolysis, which lead us to suspect that their hydroxides have
an amphoteric character analogous to aluminum hydroxide.
The fact has already been mentioned that beryllium chloride in the
fused state is a poor conductor of electricity. This would seem to
indicate the existence of more definite chemical bonds than occur in
most metallic compounds. For the most part, however, the properties
of lithium and beryllium compounds, like those of most metals, are
due to the formation of ions, through the complete removal of the
valence electrons.
Boron.
There is a wide gulf separating the hown from the aluminum
compounds, and boron is usually dassified as a non-metallic element.
This does not refer primarily to the properties of elementary boron
which, like carbon, has one form that behaves to a small degree as a
metallic conductor of electricity, but rather to the fact that boron in
its compounds does not part with its yalence electrons. Thus the tri-
halides of boron have very little of the character of electrolytes. We
have seen how the molecules of a substance like boron trichloride com-
plete the octet about the boron atom by combination) either with one
another or with other molecules.
In our present discussion of the elements with kernels of the helium
type we shall find a number of compounds whose structure can not
yet be definitely ascertained, but which seem to depart from the usual
rules of chemical combination. Of these there are none which arc
quite so mysterious as the two hydroborons B2Ho and B.H 10 ' Using
the old valence scheme one would have been tempted to write the two
formul",
H H H H H H
I I I I I I
H-B-B-H and H-B-B-B-B-H
I I I I
J J H H H H
but there are not enough valence electrons to furnish all these bonds.
Thus if we attempt to write the formula of B2HO' the number of elec-
trons available permit only such a formula as
~ Dr. N. V. Sidgwick has kindly called my attentio~ to the fac~ tha~ the i~ter
estmg beryl]jum salts of the type Bf40A,. (where A IS an orgamc aCld radltal)
may be fully explained as coot_:dinated compounds in which beryllium has the
coordination number of 4. See his note in "Nature," June 16, 1923.
H H
H:B B:H ,
H H
and there is no apparent bond between the two borons. The sugges-
tion of Eastman, which has been discussed in Chapter VII, affords
the only explanation of this molecule which has thus far been advanced.
Whatever may be the origin of the bond which joins the two halves
of the molecule together, the bond is presumably a very weak one, for
the corres{X)nrling alkyl compounds have the formula .BRa. according
to the experiments of Stock. We are reminded of the similar case
of ethane and hexaphenyJ ethane, the latter of which breaks into two
odd molecules. It is to be remarked that no bond of the hydroboron
type has been found in any other compound, nor has the substance
BH. been obtained.
Nitrogen and Carbon.
The two features which chiefly distinguish the compounds of carbon
and nitrogen from those of similar atoms with larger kerneJs are
(I) the tightness and inertness of the bonds that they form, and
(2) their ability to employ what we know as multiple bonds. The
first characteristic permits the building up of complicated structures
which are thermodynamically unstable, but which are for the most part
very unreactive. The second characteristic is evidence of an ability
to produce moderately stable electron arrangements by methods which
are not available to atoms of large kernel. While it may possibly
be des.irab1e to assume occasionally the formation of multiple bonds
in the case of elements other than those of the first period of eight,
we may be sure that the idea of the multiple bond would never have
been invented were it not for our study of the elements with kernels
of the helium type.
Since multiply-bonded compounds are classed as unsaturated we
sometimes speak of the mUltiple bond as the cause of ~nsaturation,
but this seems to me to be the very opposite of the correct view. In
the early stage of valence theory, when the single bond was the only
means known of uniting two atoms, the unsaturated compounds were
substances which could not be represented by the otdinary valence
formul"" and which might therefore have been expected to be far
more abnormal in their behavior than was found to be the case. The
mUltiple bond was invented to explain the fact that the properties of
these substances did not indicate as high a degree of unsaturation as
would be expected.
If the double bond had not been invented, the only ways in which
we could represent the structure of ethylene would be the two following :
HH HH
(A) H:¢:~:H, (.B) H:C:C,H.
COMPOUNDS OF ELEMENTJ WITH SMALL KERNELS 125
The tirst of these, having mrodd numher of electrons on each carbon

atom, should have the properties which we have found to be associated
with odd molecules. A substance of such a structure would presumably
exhibit color, form associated molecules, and be far more reactive
than ethylene in fact is. The second formula would represent a sub~
stance which would be not only unsymmetrical but also highly polarized,
the right-hand portion being charged negatively and the left-hand
part charged positively.
There may indeed be some dissymmetry in the normal molecule of
ethylene since we shall see that when two atoms, each characterized
by unsaturation, are adjacent to one another, there may be a tendency
for one of the atoms to assume a very stable state while the other
bears nearly the whole brunt of the unsaturation.
While it must be agreed that neither of the above formul", A and B
adequately represents the properties of ethylene, the same may be said
also of the ordinary double bond formula, unless it is carefully ex-
plained that the double hond implies properties very different from
those implied by two single bonds. If then we write the formula
H H
(C) H:C::C:H.
we may say that the actual behavior of ethylene corresponds to some-
thing intermediate between this and the other two formulre. Thus we
may consider that, if a molecule of the form of A should at some
moment be produced, the two odd electrons would approach and to
some extent would conjugate with one another. But this conjugation
would be by no means as complete as that which occurs when two
separate odd molecules combine, with a conjugation of their two odd
electrons to constitute a. single oond.
The exact character of this partial conjugation or partial mutual
saturation of the two atoms can at present only be guessed. In an
atom with a large kernel, and consequently a large valence shell, in
which the several electron pairs are separated from one another by
relatively great distances, we might expect that the formation of a
double bond would cause considerable distortion of the atoms con-
cerned. On ·the other hand, when the nucleus is rudimentary, as is
the case with the elements we are now considering, we seem justified
in concluding that the octet requirements of two atoms can be met by
their sharing to some extent more than one pair of electrons, and
that this sharing of electron pairs occurs without very great distortion
of the stable structures.
It has been suggested by Latimer and Rodebush that these atoms
of small kernel can be satisfied with something less than the normal
octet, namely, by a group of four, or especially by a group of six,
but this does not seem to he a complete solution of the problem, for
if such were the case We should expect to tind monatomic carbon with
its quartet of electrons, and monatomic oxygen with its sextet of
electrons, to be far more stable than they are in fact.
We arc therefore led to the conclusion that the properties of sub-

stances with multjple bonds are due to some sharing of two or three
electron pairs, although this shar;ng js probably less complete than
that indicated by the usual graphical formula with two or three bonds.
Nevertheless the process must be assumed to have some physical reality,
in order to account for the existence of those isomers which depend
upon the failure of free rotation about the double bond.
Also for acetylene we might write three types of formula,
H
(A) H:C:C:H, (B) H:C:C:, (C) H:C:::C:H
The formula 13 suggests one employed by Ncf. Now sjnce acetylene
acts as a weak acid. the two hydrogen atoms may he rega,rded as mobile,
and the two forms A and Ii would give the same ions, namely,
\Ve may therefore regard the forms A and B as belonging to that

simple class of tautomers which are obtained merely by the transfer of
hydrogen from one lone pair to another. On the other hand the alkyl
derivatives should yield isomers. and the fact that such isomers have
not been isolated is an argument against the structures A and B.
However, the formula C does not express the properties of acetylene
without a good deal of explanatiun. Here again the behavior of acety-
lene seems to he intermediate between that which would be represented
by the formula C and that expressed by one of the other formul",.
Indeed if. in formula A, we consider a pair of electrons from each
atom drawn toward the central bond, it would be difficult to state the
exact point at which the single bond would 4_e converted into a triple
bond. Of course it will be realized that in all these discussions our
standardized formul<e do not very adequately represent the three-
dimensional arrangement of the electrons in space.
We have already commented upon the fact that the unsaturation
of a compound containing a triple bond is less, as shown by aU of
our best criteria, than the unsaturation of a compound with a double
bond. This is entirely at variance with the strain theory of Baeyer,
and we are at a loss to account for it except perhaps by the assump-
tion that the two types of molecules do in some measure approach the
two symmetrical forms represented by the formul:e A for ethylene
and acetylene. In the former case the process is attended by the
division of an electron pair; in the second case it is not. It is not easy
to avoid the suspicion that this fact is connected with the greater unsat-
uration of the double bond.
A discussion of the triple bond leads us to a consideration of the
interesting properties of elementary nitrogen which has generally been
assigned the formula N "" N. This substance is distingnished by its
very great lack of reactivity. It is aJso stable in a thermodynamic
sense with respect to a great number of its compounds, which ten<l
COMPOUNDS OF ELEMENTS WITH SMALL KERNELS 127
to decompose in such manner as to evolve N 2- For these reasons

Kossel (1916) remarked: uN, is known to be very inert chemically.
It is very little disposed to take on foreign electrons, and while it
cannot be considered fully comparable with the noble gases, still it
can for our purpose pass as a very stable structure. If one nitrogen
is replaced by a carbon, a structure is obtained by which the positive
charge is smaller by one unit, but also has one less electron than in
the stable N,. It behaves therefore with respect to N, just as a
halogen, according to our assumption, behaves with respect to a noble
gas. Thus CN functions chemically as a halogen."
This idea has been extended by Langmuir, who has pointed out
the great similarity in physical properties between nitrogen and carbon
monoxide. Each of these molecules has the same number of electrons
and the same molecular weight, and it is therefore perhaps not sur·
prising that in their boiling point and other similar properties they
are closely analogous. Mr. Urey has called my attention to the fact
that the distance from nucleus to nucleus, as determined by means of
a study of the ultrared absorption bands, is the same for both sub-
stances, namely 1.14 x 10-8 em.
It is possible, however. to exaggerate the similarity between the
two gases. In their chemical properties the resemblance disappears.
Carbon monoxide forms many addition compounds as in the metal
carbonyls, the complexes with cuprous salts and with haemoglobin.
However, if we are to assume precisely the same electron structure
for N, and for CO, the latter must be electrically polarized in the
sense that the oxygen is positive and the carbon negative. since in
the neutral state the oxygen atom has one more electron than the
nitrogen atom, while the carbon atom has one less. Such polarization
would probably suffice to explain the difference in chemical behavior.
On the other hand if the electrons in carbon monoxide are shifted in
such manner as to diminish the degree of polarization of the molecule,
the distortion of valence shells so produced might equally well account
for the chemical reactivity of carbon monoxide.
. The particular structure which Langmuir proposed for nitrogen and
carbon monoxide, and which had previously been suggested to me by
Professors Bray and Branch, I do not regard as probable. Langmuir
suggested what is essentially a quadruple bond, such that two atomic
kernels lie together inside of a single octet. There seems to be nothing
in the properties of these substances to necessitate such an ad hoc
assumption.
On the other hand we have much reason for believing with Kassel
and Langmuir that there are substances like cyanide ion and carbon
monoxide which have essentially the same electronic structure as
nitrogen. To these we may add also the acetylide ion, and, with the
reservations regarding the meaning of the triple bond which we have
already made, we may assign to these substances the formulre
:N:::N:, :C:::O:, [:C:::N:]-, [:C:::C:]--.
l:z8 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
In accordance with this view we see that the ion CN- is the same
whether it comes from a cyanide or from an isocyanide. By adding
hydrogen ion to a lone pair of the carbon or of the nitrogen we ob-
tain hydrogen cyanide or hydrogen isocyanide, and these two sub-
stances therefore represent the same kind of tautomerism which we
have so frequently mentioned. By employing alkyl groups in place
of hydrogen the corresponding isomers may be obtained.
\\rhcn we examine these four substances which have presumably
approxlmately the same electronic structure we note that the last two
are ions, that carbon monoxide on account of the difference in charge
of the kernels must be polarjzed, but that nitrogen is neither polarized
nor is it an ion, and if we bear in mind the unexpected lack of un-
saturation which we have seen to be associated with the triple bond,
and consider also the remarkable firmness which in general char-
acterlzes the bonds of nitrogen, we have, I believe, a sufficient ex-
planation of the peculiar inertness of elementary nitrogen.
We may recall that Nef ascrihed the isocyallides and acetylene to
the class of compounds with bivalent carbon. If we interpret this
statement as meaning merely that the carbon is in the same kind of
structure as it is in carbon monoxide, we are now in a position to
agree with his conclusions.
The oxides of nitrogen form an interesting and unusual group
of compounds. Nitric oxide. NO, we have already discussed as onc
of the substances which have odd molecules. It has eleven valence
electrons; one more than the molecule of nitrogen. Of all the sub-
stances which have odd molecules it exhibits the least degree of un-
saturation. It is colorless, and at ordinary temperatures does not
associate into double molecules, although this process seems to occur
at low temperatures. ,By some method which is not yet understood,
the odd electron is obviously much more firmly held in the molecule,
and the molecule itself is more nearly saturated, than in the case of
any other odd molecule.
Professor Branch has pointed out to me that this behavior of nitric
oxide is very closely related to the anomalous properties of the nitroso
compounds. When NO combines with another free radical (odd
mo1ecule) such as an alkyl group, we should expect this union of two
odd molecules to produce a fully saturated compound. But this is not
the case. Not only does NO in the free state behave nearly like a
satnrated substance, but when combined with an odd molecule like
methyl it does not appear to conjugate with the odd electron of the
latter. So the resulting compound, although possessing an even number
of electrons~ has the properties of odd molecules. The nitroso com~
pounds in general are highly colored, and they almost invariably tend
to form double molecules, as though each single molecule had an odd
electron...
The first stage in the reaction between nitric oxide and oxygen
appears to be as follows:
NO+O.=NO, .
COMPOUNDS OF ELEMENTS WITH SM.4LL KERNELS 12<}
The assumption that this reaction is reversible at ordinary tempera-

tures, and proceeds to a' very slight extent at high temperatures, ac-
counts for the very interesting observation that the reaction hetween
nitric oxide and oxygen, to give NO:;! or N 20'" diminishes in ~peed with
increasing temperature. The intermediate substance NOs. which is
assumed for this reaction, can actually be isolated at low temperatures.
It represents another type of odd molecule; it is intensely colored, and
presumably in other ways is more unsaturated than nitric oxide.
Nitrogen dioxide is another suhstance with an uneven number of
electrons, and it shows all the characteristic properties of the odd mole-
cules. I believe that we are hardly in a position as yet to give definite
electron structures for these odd molecules. Presumably if there is
any part of the molecule which is a seat of ullsatunl.tion the odd
electron seeks that position, and to some extent conjugates in such
manner as to reduce as far as possible the total unsaturation.
The three remaining oxides of nitrogen, NzO, N 2 0 a and
N 20 5 , may be represented as mUltiply-bonded or hy the formula!,
:0: :0:
'():N:(j:N:ci:
.. .. ..
Oxygen ang Fluorine.
Oxygen and fluorine are the two most electronegative elements.
By this we mean that they are the elements which sbow the strongest
tendency to complete their octets and to obtain as nearly as possible the
exclusive possession of the eight electrons. They therefore either
take electrons completely away from other atoms, forming ions, or
they draw electron pairs away from other atomic centers, possibly into
a secondary valence shell.
For this reason it seems at first sight surprising that water and
hydrofluoric acid are weak electrolytes. This, however, is not an
isolated phenomenon. Hydrofluoric acid in aqueous solution is weaker
than hydrochloric. acid, water is a weaker add than hydrogen sulfide,
which in turn is weaker than hydrogen selenide and hydrogen telluride,
while the same phenomenon appears also in the nitrogen group. It
is evident that hydrogen ion, when it adds to a lone pair of oxygen
or fluorine, forms a firmer bond than when it combines with other
lone pairs, and this it can do, probably because of its small size,
without any great defonnation of the normal octets of oxygen and
fluorine atoms. The hydrogen is certainly held very close to the other
atom; we have seen that the distance separating the two atoms in the
molecule of 'hydrofluor~c acid gas is probably less than I x IO~S em.
The firmness with which the ions of these elements of smaIl kernel
hold hydrogen ion is further illustrated by the fact that neutral com-
pounds. like ammonia and water have a far greater tendency to add
another hydrogen ion, forming onium compounds, than the correspond-
ing substances, phosphine and hydrogen sulphide, respectively.
As far as we know, oxygen and fluorine do not furnish the centra!

atoms of anions as so many other elements do. However the problem
of the structure of ozone is a very interesting one in this connection.
Ozone is usually given a ring formula which we may readily translate
into an electron structure, but it has seemed to me that it may possibly
have the form
" .. ,.
:0: :O:?:, or :9:0:?:
......
which would make it the al1alog-uc of sulfur dioxide. 1 There is some
intimation in the literature that ozone is absorbed by basic substances
without decomposition, and if so we should regard it as the anhydride
of an acid resembling sulfurous acid. It would be interesting to
obtain further experimental evidence. If this view of the structure
of ozone is correct, the fact that fluorine reacts with water to produce
ozone might lead us to suspect that in the reaction there are inter-
mediate substances, which would resemble the oxyacids of chlorine,
but in which the oxygen would be taking the place of the central
chlorine, while ftuorlne would occupy the outer positions which are
held by oxygen in the other halogen acids. It is possible that experi-
ments at low temperatures might possibly lead to the isolation of such
compounds between fluorine and oxygen.
The question which we have just raised regarding the constitution
of ozone is of far greater consequence than the question of the con-
stitution of hydrogen peroxide which is more frequently argued. The
two formulce for ozone, the ring formula and the one of the type of
SO" are radically different in principle. On the other hand, the two
formul", which have been given to hydrogen peroxide, namely HOOH
and H,OO are to all intents and purposes identical. Hydrogen peroxide
is an acid whose ion is expressed by the formula
[ :0:0:]--
.. - ' . '
Whether the two mobile hydrogen atoms go to the same or different

oxygen atoms is unimportant. This is the same kind of tautomerism
which we have mentioned in the case of sulfuric acid, phosphoric acid,
hydrocyanic acid, acetylene and hydroxylamine.
j I find that a similar suggestion has just been made by Lowry, Trans. Faraday
Soc., IS, Part 3, page 3, 1923.

Chapter XI.
Elements in Positive and Negative States.
When we speak of a negative element we can only mean an ele-
ment which in a neutral state will take on electrons to form stable pairs
and octets. \Ve. cannot assume that a neutral atom of chlorine exerts
any appreciable electrostatic force upon an electron at a distance.
We ran only say that when an electron is added to a chlorine atom,
in such a way as to complete its group of eight, a very stable system
is produced. \Vhen chloride ion is thus formed this ion wlll not
attract but wi]} repel negative electricity.
Our eXlsting nomenclature is extremely misleading. VJ" e say that
chlorine is a negative element because it tends to take up an electro!?
and we also say that in sodium chloride we have negative chlorine, be-
cause it has taken up an electron and has no tendency to acquire any
marc. \Vhen an atom of chlorine instead of gaining an electron has
lost one, \ve say that we have positive chlorine, although the atom then
has a great tendency to take on two electrons. I shall not, however,
attempt to offer a substitute for these time-honored usages which are
probably so well understood that they will cause no serious confusion.
Perhaps it will somewhat clarify the situation if we agree that an
element which tends to take up electrons is a negative element, that
when it has taken up a sufficient number of electrons it is in a negative
state, and that when its need for electrons is not fully satisfied it is
in a positive state.
That the process of forming negative ions is very different from
that which was assumed in the electrochemical theory is shown by the
fact that a negative element, if it cannot acquire its group of eight,
may often give up an electron in order at least to possess an even
number of electrons in its valence shell. Thus the conductivity of pure
liquid iodine indicates that one atom of iodine has acquired an electron
to form iodide ion, while another has parted with an electron to be-
come iodous ion with a group of six electrons. As a rule, however,
both atoms complete their groups of eight by sharing an electron pair,
The Problem of Electromers.

Hypoiodous acid is a very weak add indeed. In its aqueous solu-
tion it does not increase greatly the hydrogen ion concentration due to
pure water. It probably is amphoteric and dissociates to a small de-
gree into the ions I' and OH-. So the compound leI may be regarded
131
as a very weak salt, iodous chloride. Such considerations have led

the proponents of the modern dualistic theory to the conclusion that
all compounds of iodine can be regarded as containing either negative
iodine, as represented by 1-, or positive iodine, as represented by r.
We have seen how untenable such a hypothesis is, and that it is neces..
sary to regard the highly polar molecules as extreme examples of the
displacement of the pair of electrons which constitutes the chemical
bond.
There can be no question but that iodine in hypoiodous acid is in
a very different state from iodine in potassium iodide, and there is
no objection to sayjng that it is more positive in the former sub-
stance 1han in the latter but we could exhibit a whole series of com-
f
pounds which arc intermediate between these types, and a classification

of these substances into two distinct groups, one containing positive
iodine and the other cot1tainillg nef,rative iodine, would be wholly
arbitrary.
We 110 longer therefore see any justification for the assumption that
nitrogen trichloride must be either N~-HCI-31 or N---Cl+ a • Such a
pair of hypothetical substances ha\'e become known in the modern
dualistic theory as electromers,-two substances which are supposed
to have the same atoms in the same arrangement but with a different
distribution of electronic charges. If we admit that the nitrog.en in
nitrogen trichloride is nearer to the state of nitrogen in ammonia
than it is to nitrogen in nitrous acid, we no longer imply that we
must give it the second of the above formuIre, or that it would be
possible to obtain another kind of nitrogen trichloride corresponding
to the lirst formula.
Nevertheless if we apply the quantum theory in any thoroughgoing
manner to chemical combination we cannot regard the pair of electrons
which serves as a bond between hydrogen and chlorine as capable of
being moved in a continuous way from the one atom to the other.
Rather we are led to assume that such a pair must occupy one of
the finite though possibly large number of definite positions between
the two atoms, and two molecules holding the bonding pair in differ-
ent positions (or energy leyels) might be called eleetromers. Thus
an atom of hydrogen with its electron in the lirst energy level might
be called an electromer of an atom which for a brief space of time
holds the electron in one of the other energy levels. However, even
if we assume the existence of isomeric molecules it is hardly to be
expected that we shall be able to segregate and isolate the several
species of molecules so as to obtain electromeric substances.
• Compounds Between Negative Elements.
A negative element not only t~nds to acquire electrons to fonn

its stable group, but also to obtain exclusive possession of the electrons
which make up its octet. In substances of the extremely polar type.
such as calcium oxide, the oxygen may be considered to be in a state
ELEMENTS IN POSITIVE AND NEGATIVE STATES 133
of oxide ion, and thus to have acquired complete and sole possession
of its group of eight electrons. But in the majority of oxygen com-
pounds in which the atoms of this element are attached by chemical
bonds, one or more of its four pairs of electrons is shared with
another atom, and this always produces some distortion of the valence
shell. In all such cases the oxygen must be regarded as in a more
positive state than it is in oxide ion.
Especially when two electronegative atoms are combined we may
consider that a state of tension exists, in which each atom, so to
speak, strives to obtain sole possession of the bonding pair. This
pair is therefore drawn away from each atom more than it would be
in the 1110st stable form of the octet.
In such cases it has been customary in certain quarters to state
that one of the atoms is in a positive condition. I would amend this
by saying that whenever two electronegative atoms are united by a
chemical bond both atoms arc in a morc positive condition than when
they are combined with electropositive elements or are in the state
of their normal ions.
As examples of cotu~unds of this type we may consider Cl- Cl,
Cl- OH, CI- NH" HO - OH, H,N - OB, B,N - NB" NCI, and
other similar compounds. In all such cases those who have adopted
the modern dualistic theory have attempted to decide which of the
atoms is positive and which negative. It is said that chlorine is
positive in CIOB. and there is perhaps some. jusification for considering
that in this substance chlorine is a litt1e farther from the state of
chloride ion than the oxygen is from the state of oxide jon, but it
is certainly not so positive as iodine in IOH. After aU, the main
thing in aJl such cases is to recognize that both of the atoms con-
cefned are in the state that we call positive, and that a state of tension
exists that can best be relieved by breaking the bond between the
two atoms and attaching to each atom a more naturally electropositive
element or radical. Such a process would be classed as a reduction,
without the necessity of telling which atom is reduced.
Hitherto little has been said regarding the actual distribution of
electricity within a molecule, nor would it be wise in the present state
of our knowledge to attempt to speak with any definiteness in this
regard. Howeyer, there are certain elementary deductions which may
be safely drawn from our theory.
When two atoms of chlorine combine to form the molecule Cl:h
each atom contributes half of the bonding pair, and it is to be pre-
sumed that in the most stable state of the molecule the whole struc-
ture is symmetrical with respect to this bonding pair, We might then
say that the pair belongs equally to the two atoms and therefore that
it furnishes to each atom the equivalent of one electron, which, together
with the six electrons which each atom holds by itself, makes the
seven of the neutral chlorine atom.
In a purely formal way we could in other cases assign to an atom
one electron from each of its bonds and thus obtain a value for the
'34 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
electric charge of each atom. .But this would be an entirely arbitrary

proceeding, taking no cognizance of the shifting of electrons which
we must always consider when the two sides of the bond are not
identical, and which undoubtedly occurs in some molecules even of a
symmetrical type, due to thermal agitation.
Even this formal method of assigning electrons to the indivi4ual
atoms would result often in ascribing charges to certain atoms. Thus
in the numerous cases in which oxygen has hut a single bond, as
for example in the amine oxides,
R
R:N.:~: ,
R
the oxygen atom furnishes neither of the electrons which constitute
the bond. This atom has sole possession of the six electrons which
offset the six positive charges of its kernel, and since it also has part
possession of the bonding pair it must be considered to be in a negative
condition. Presumably such a substance as amine oxide is therefore
considerably polarized-. In the same way a neutral substance like
boron trifluoride may be considered to be made negative when it
completes its octet by employing one of the lone pairs of water or
ammonia.
In any of the numerous cases of tautomerism which consist in
the passage of hydrogen ion from one lone pair of electrons to an-
other, as in the tautomerism of hydroxyl amine, namely,
H
H:N:O:H = H:N:O:.
H " ii "
the transfer of the positively charged nucleus of hydrogen would,
formally considered, result in an increase in the positive charge of that
portion of the molecule to which it goes, and probably such a change
of polarity actually occurs, although it may be Jargely offset by some
displacement of the whole electronic structure.
When hydrogen ion adds to a lone pair of electrons to form an
onium ion we should expect from considerations of electric force alone
that the hydrogen iOIl would most readily attach itself to that portion
of a molecule which is negatively charged. We have seen that when
one hydrogen ion or other positive radical has been attached to an
atom, thus giving it a positive charge, this ordinarily prevents the
formation of a second onium bond, as in the case of substances of
the sulfoniurn type, It seems to be a universal rule that the tendency
of hydrogen ion or a (positively charged) alkyl radical to add to a lone
pair, in any part of a molecule, increases if that part is made. more
negative. and diminishes if it be made more positive.
In accordance with this rule nitrogen is less disposed to form
substances of the ammonium type when it is in a positive state than
ELEMENTS IN POSITIVE AND NEGATIVE STATES 135
when it is more negative. Thus when nitrogen is attached to chlorine,

oxygen or nitrogen, it combines less readily with hydrogen ion and is
said to be less basic. Thus chloramine is a weaker base than ammonia.
In other words, the first of the two following reactions occurs to the
least extent;
ClNH, + H' = ClNH,'
NH,+H'=NI-I,'.
Nitrogen trichloride is hardly basic at all. So also hydroxylamine
and hydrazine are very much weaker bases than ammonia.
We have seen that univalent oxygen as it exists in an amine oxide,
in spite of its attachment to another negative element, is in a nega-
tive state. It should therefore be able to add hydrogen ion. Therefore
from this standpoint also we have confirmation of the view expressed
in a previous chapter that an amine oxide adds hydrogen ion at the
oxygen.
The strain which exists in a bond between two negative elements
manifests itself in a variety of ways, and is responsible for numerous
types of decomposition or rearrangement. Let us consider once more
the compound formed when an amine oxide combines with methyl halide
to produce the ion RaNOCHa+, This ion very readily decomposes in
alkaline solution to furnish hydrogen ion and to leave formaldehyde
and free amine. It seems to me that we can form a provisional picture
of the mechanism of this decomposition. Representing the ion uy
the structure
we may consider the bonding pair between nitrogen and oxygen as

drawn away from the normal position of the palr in the oxygen octet.
This struggle for electrons would tend to draw toward the oxygen the
electrons which form the bonding pairs between carbon and hydrogenJ
so that the atoms of hydrogen would behave more like the hydrogen
of an acid than they ordinarily do in an alkyl group. So we may
consider that in the presence of an alkali one of these atoms of
hydrogen occasionally is drawn off as hydrogen ion. The pair of
electrons left free would then form a double bond with the oxygen,
thus producing formaldehyde, and the nitrogen would take sole posses-
sion of the former bonding pair, giving trialkylamine. Such a picture
of the mechanism of this decomposition seems extraordinarily satis-
factory, but in the present stage of chemistry any theory as to the
detailed mechanism of a reaction must be considered tentativeJ and
it must be admitted that there are a number of quite similar decomposi-
tions lmown to organic chemistry which apparently cannot be explained
without a far greater amount of molecular rearrangement than is
assumed in this simple case.
An extremely interesting substance which is probably closely re-

lated to amine oxide is produced when ammonia reacts with an aldehyde
to form the compound known as aldehyde-ammonia. We may pro-
visionally assume that the reaction between these substances consists
in the breaking of the double bond of the carbonyl group, in accordance
with the following scheme:
:0: H :0: H
n
R:C + :N:H R: C :N:H
H H H H
The formula of the compound as now written probably represents
the actual structure of some of the molecules, since ammonia can be
readily split off again. However, in this formula the oxygen is singly
bonded as in an amine oxide, and is therefore in a negative condi-
tion. Hence olle of the liable hydrogens should wander to the oxygen,
resulting in the molecule
H:O:
R:C:N:H
HH
This is probably the predominant form and accounts for the carbinol
(slightly basic) properties of the aldehyde-ammonia. If in some
molecules two hydrogens wander to the oxygen, we have a structure
from which water could readily be split off, leaving the imide. This
reaction also is characteristic of this class of substances.
By our definition of negative elements as those which show a
strong tendency to acquire electrons to fom1 a stable group, it is
evident that we must regard hydrogen as a negative element. It js
true, hydrogen rarely takes exclusive possession of a bonding pair
tD ifxm nYGrUk it.."\"}, iVlt it seems to exerc~se a ver), considerable puJJ
upon a bonding pair, especially in sOl1).e inorganic acids, as we shall
see in the next chapter.
Chapter XII.
Remnants of the Electrochemical Theory.
In our development of the new theory of valence we have found
it possible to agree with ,,'erner in many points. As the text of the
present chapter we might use his statement: "The electrochemical
phenomenon accompanying the saturation of principal valences is a sec-
ondary one, and quite different from the purely chemical one. It
may accompany chemical change but is not a necessary consequence."
There are, however, numerous reactions the trend of which is largely
determined by conditions of electric polarization within the molecule,
and to these we shall now give our attention.
When a typical salt is dissolved in a solvent which is not highly
polar-for example in a solvent of low dielectric constant-it often
shows Httle evidence of electrolytic dissocia6on. In such cases the
undissociated molecule may be regarded as held together by the electric
forces between its oppositely charged parts. This is perhaps the only
type of molecule in which chemical combination is the result of purely
electrostatic forces, in the ordinary sense. Nevertheless there are many
other cases in which the union of two chemical species is due to the
formation of a chemical bond, but in which the extent to which the
union occurs is largely determined by electrostatic forces.
We have seen that an element which is naturally quadrivalent does
not satisfy its valence of four through the formation of an onium
complex, if this requires the addition of hydrogen ion to an already
positively charged ion. On the other hand, the positive hydrogen ion
does not readily dissociate from an ion which is already negative.
Thus hydrogen sulfide dissociates according to the equation H,S =
H+ + HS- and the dissociation constant is 10-1, but we find a far
lower dissociation constant, about 10-11'; for the ionization HS- ==
H> + S- -. A similar example is furnished by the two reactions,
NH,-+H>=NH, ,
NH, +H>=NH? .
While the second of these reactions is easily reversible in aqueous
solutions, the first runs so nearly to an end that an amide, even when
dissolved in a very alkaline solution, is completely hydrolyzed, as far
as we know.
Since it is in reactions which involve the union or dissociation of
ions that we might expect the clearest evidence of the influence of
electrostatic forces, it is fortunate that we possess a large amount of
'37
IJ8 VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
information concerning the constants of dissociation of weak acids and

bases.
The Strength of Acids and Bases.
We arc so habituated to the use of water as a solvent, and our
data are so largely limited to those obtained in aq~eous solutions, that
we frequently define an acid or a base as a substance whose aqueous
solution gives, respectively, a higher concentration of hydrogen ion
or of hydroxide ion than that furnished by pure water. This is a
very one-sided definition, but it will suffice for the moment while we
consider those substances whose acid or basic properties are due to the
presence of the hydroxyl radical.
Let us consider a substance whose structure and modes of dissocia-
tion are represented by the following schemes,
:~:9:H = [:~T + [:?,:HT
::i\,:S?:H = [:~:¢:l + W
If the dissociation otcurs chiefly by the first method, the substance
is called a base; if chiefly by the second method it is called an acid.
Frequently both dissociations occur, and the substance is said to be
amphoteric. Since the product of the concentration of hydrogen and
hydroxide ions is limited by the dissociation constant of water, a sub-
stance in aqueous solution cannot be at the same time a strong acid and
a strong base.
There seems, however, to be a far more fundamental opposition
between acid and base than the one we have just mentioned. We fre-
quently deal with a substance like alcohol which is so weak an acid
and so weak a base that the limitation of concentration of hydrogen
and hydroxide ion is of no importance, and yet in stIch a case we shaH
probably find that almost any change in the molecule which makes it a
stronger acid makes it a weaker base, and vice versa.
Reverting to our type substance XOH, X may be an element or
radical which may without difficulty lose electrons. In such case oxygen
which strives for exclusive possession of its group of eight will break
the bond with X and take possession of the bonding pair, thus forming
X' and OH-. On the other hand. if X is an element or radical which
exerts a strong pull upon the bonding pair, all of our evidence indicates
that this produces an efIe<:t upon every one of the pairs in the oxygen
octet, which may be regarded as a displacement toward the left. Thus
the tightening of the bond between X and 0 loosens the bond between
o and H, and the displacement of the electrons away from the hydrogen
makes the la~r more free to assume the condition of hydrogen ion.
If we consider coi-responding hydroxides of nitrogen, phosphorus,
arsenic, antimony and bismuth, we see~ the effect. of the diminishing
pull of electrons by the central atom, as we proceed from nitrogen to
bismuth. The hydroxides become progressively weaker acids and
stronger bases. .
REMNANTS OF THE ELECTROCHEMICAL THEORY '39
Such a simple explanation a~counts satisfactorily, in the main, for

t~e observed strength of organtc acids and bases. We have already
dIscussed the case of chloracetic acid, to which we give the formula
R:O:
:d:c: C:O:R
" :it ..
The substitution of one methyl hydrogen by chlorine produces a greater
pull upon the electrons of the methyl carbon, and this causes a dis-
placement which seems to occur throughout the molecule, finally pulling
the electrons away from the hydrogen and permitting a greater dissocia-
tion of hydrogen ion. The substitution of a second and third chlorine
heightens the effect.
It is not at all certain tmt the whole effect is produced through the
carbon chain. If we knew how to construct thoroughly satisfactory
spatial models of our moleCllles, we might see how in certain cases
there might be some more immediate steric effect which would exag~
gerate the effett produced through the chain. But that the latter effect,
passing through the molecule atom by atom, is a real one we have no
reason to doubt. One substitution of hydrogen by chlorine in propionic
acid gives an acid of about the same strength as monochloracetic, when
it is an alpha~hydrogen that is replaced: hut ~-chlorpropionic acid is
much weaker. Still less marked results of the chlorine substitution are
to be seen in y-chlorbutyric and in b-chlor\'aleric acids, the latter of
which is hardly stronger than valerie add itself.
Other atoms or radicals which exert a similar pull upon the electrons
give entirely parallel results when substituted for hydrogen. An
apparent anomaly was found in the case of the amino-adds. The group
NH2 is regarded as a highly negative radical. but the amino-acids show
very low electrical conductivity. This phenomenon is explained in
an entirely satisfactory manner as due to the formation of inner salts.
In a paper by E. Q. Adams (1916) in which he makes clear a number
of important points regarding the strength of acids, it is shown that
under similar circumstances hydrogen ion dissociates from the carboxyl
group to a greater extent in rnonoaminoacetic add than in mono·
chloracetic acid. But the hydrogen ion which is set free recombines,
forming an ammonium complex with the nitrogen. There are some
cases in which hydroxyl acids have a lower conductivity than might be
expected, and here too we may suspect that the hydroget; ion, which
comes from the carboxyl, fonns to a slight degree an oxomum complex
with the hydroxyl group.
The dibasic acids make an interesting study. There has been some
confusion existing as to the meaning of the first and second dissocia...
tion constants. It has occasionally heen supposed that these two con-
stants are indicative of a speciaJ difference between the hydrogens oE
the two carboxyls, hnt this is hy no means the case. In a symmetrical
dihasic acid tbe two carboxyl hydrogens may be regarded as identical
in pruperties and the chance of dissociation is the same for one as for
the other. If the dissociation of one of these hydrogens does not affect
the chance of dissociation of the other hydrogen. as it may not if
the two carboxyls are very far removed from one another, then
it has been shown by E. Q. Adams that the second dissociation constant
of the acid will be just one-fourth of the first dissociation constant.
He found in fact one or two cases in which the carboxyl groups were
so completely removed from one another that this proved to be the case.
Ordinarily, however, the two carboxyl groups profoundly influence
one another. and since carboxyl is a negative group and exerts a pull
upon electrons, the first dissociation constant of a dibasic acid is usually
much higher than the dissodation constant of a corresponding mono-
basic acid. This effect is the more marked the nearer the two carboxyls
are to one another. Thus in the series of compounds with varying
numhers of carbon atoms in a chain, and with the two carboxyls at
either end of the chain, the first dissociation constant varies from 10-(1;)
for a chain of ten carbon atoms to 10-1 for a chain of two carbon atoms
(oxalic acid).
V/hen, however, a carboxyl group has given off a hydrogen ion and
has a negati've charg-e, it no longer exerts a pull upon neighhoring
electrons, but behaves rather as a strongly positive group. ~\nd thus
we find, in the series of compounds which has just been mentioned,
that the ratio of the second dissociation constant to the first is Yro in
the case of the ten-carbon chain. and Y2000 in the case of oxalic add.
The case of oxalic acid is therefore like that of hydrogen sulfide. but
in the latter the discrepancy between the first and second dissociation
constants is far greater still.
The same ideas which are useful in the interpretation of the dis-
sociation of the weak organic acids are equally applicable to inorganic
adds. Here the problem has been very ably discussed hy Latimer and
Rodebush. The three acids 01 phosphorus. namely. hypophosphorons
acid, phosphorous add, and phosphoric acid. they agree with \'Verner
in writing
H :0: :0:
:O:P":O:H H:():P:O:H H:O:P:O:H
.. Ii' H
.. :g: ..
H
From the fact that these three acids are of approximately equal strength
they conclude that hydrogen when attached to phosphorus behaves as
a decidedly negative element and exercises at least as great a pull as
the hydroxide mdic.l. therefore causing approximately the same effect.
On the other hand. they believe, contrary to the views of Werner,
that sulfurous add is chiefly composed of ·molecules of the form
H:O:S:O:H
·':0:·'
REMNANTS OF THE ELECTROCHEMICAL THEORY '4'
In this formula sulfur possesses a lone pair of electrons, and this pair,
instead of exerting a puU which would distort the oxygen octet, is
mobile and tends to prevent such distortion. It hehaves therefore like
a positive group, and sulfurous acid is a weak acid. This distinction
between a lone pair of electrons and a pair of electrons which, being
attached to a negative atom, is therefore pulled away from the
atom, accounts in a very simple manner for the great difference between
the acids of elements in their highest state of oxidation, which have no
lone pairs on their central atoms, and which are strong electrolytes,
and the acids of elements in their lower stages of oxidation, which
possess such lone pairs and are usually weak electrolytes. (The next
to the highest acids of the halogens. such as chlotic acid. seem to con-
stitute an exception to the general rule, for which at present no explana-
tion has been offered.) If we choose to write the formula of nitrous
acid using no double bond, and assigning to nitrogen a sextet of
electrons, which at least partially represents the facts, namely,
:~:~:?:H ,
we see that there are two possible explanations of the weakness of
nitrous acid. In the first place, the lone pair on the nitrogen atom
prevents that distortion of the oxygen octet which tends to set free
hydrogen ion, and in the second place any hydrogen ion which does
dissociate may in part be attached to the lone pair of the nitrogen, just
as we have seen that the hydrogen from an amino-acid recomhines
with the nitrogen. It is evident that whether we write the formula
with hydrogen attached to nitrogen or to oxygen, the ion N02 - is the
same, and the tautomerism is of the sort that we have so often con-
sidered. Hydrogen ion will ulldoubtedly attach itself in the main to
that atom with which it forms the firmest bond. and therefore in
general, when there is more than one possible point of union, the mole-
cule which corresponds to the weakest acid will be formed.
The Definition of Acids and Bases.

When we discuss aqueous solutions of substances which do not
contain hydroxyl, it is simplest to define a base as a suhstance which
adds hydrogen ion. Thus ammonia adds hydrogen ion to form
ammonium ion, and the degree to which this occurs will vary as we
substitute other radicals for hydrogen. Indeed if we wish, we may
consider ammonium ion as an acid and say that its strength as an acid
is increased when hydrogen is replaced by a or OH or NH,. This
is precisely the same as saying that ammonia is a weaker base when
such substitutions are made. So we might go through a long list of
organic bases and show, just as in t~e case of !he aci~s, how. tt;e. pull
exercised .upon electrons by the varIOus negatIVe radIcals dl1rumshes
their ability to attach hydrogen ion.
Since hydrogen is a constituent of most of our electrolytic solvents,
the definition of an acid or base as a substance which gives up or takes
'4Z VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
up hydrogen ion would be more general than the one that we used
before, hut it will not be universal. Another definition of acid and
base in any given solvent would be the following: An acid is a sub-
stance which gives off the cation ()r combines with the anion of the
solvent; a base is a substance which gives off the anion or combines
with the cation of the solvent. So potassium amide is a base in
ammonia, while potassium chloride would likewise be called a base
in liquid hydrochloric acid.
Even this very broad definition is not entirely satisfactory. We·
are inclined to think of substances as possessing acid or basic prop-
erties, without having a particular solvent in mind. It seems to me that
with complete generality we may say that a basic substance is one
'll}hich has a lone pair of clcctroll-S which may be used to cMnplcfc the
stable group of altotlter atom, and that a1l acid substance is one which
can cmplo), a (aile pair fron~ another nt.o{ecu[e in completing the stable
group of one of its own atoms. In other words, the basic: substance
furnishes a palr of electrons for a chemlcal bond, the acid substance
accepts such a pair.
In this sense all substances which have lone pairs of electrons,
capahle of employment in onium formation, are basic substances. On
the other hand. substances like hydrogen lon. iodous ion, silicon
dioxide, sulfur trioxide and boron trichloride are acid substances, since
hydrogen ion will accept a pair of electrons to form its stable grou):
of two. and the remaining substal1('"es will accept pairs of electrons tc
complete their stable groups of eight. Some of these substances whid
we hayc set down as acid are ohviouslv basic as well; the sulfur aton
in sulfur trioxide is add) but the oxygen atoms in sulfur trioxide ma~
act in a basic manner.
Other Factors Determining Dissociation.

We are not in a position as yet to give any quantitative measure t
the pull that any given atom or radical exerts upon the electrons, an
this is partly due to the fact that there are other factors which hel
in determining the strength of weak electrolytes. Nevertheless it
of interest to attempt in a rough way to arrange radicals accordir
to their electronegative or electropositive character. Thus we see
to be justified in saying that hydrogen is more negative tban a meth
group but more positive than a phenyl group. Methyl alcohol is
weaker add than water, phenol is a stronger acid; on the other hat
methylamine is a somewhat stronger base than ammonia, while anm·
is a very much weaker base.
It is, however, necessary to exercise some caution in making the
deductions. 'The electrolytic dissociation is important as an index
the pull exerted upOn electrons by the various types of atoms, l
there are doubtless a number of independent factors which help
determine the degree of dissociation of a weak· electrolyte. For (
ample, the dissociation is largely determined by the extent to which (
or both of the ions of the dissolved subo;tance form complexes w
REMNANTS OF THE ELECTROCHEMICAL THEORY 143
the solvent. This is a very important question when we are considering

:' given electrolyte in various solvents. but the extent to which hydrogen
10n combines with water to form hydronium ion is not of much con-
sequence when we are comparing a number of acids, all in aqueous
solution.
Even if it is to be admitted tbat the displacement of the electronic
structure, which is assumed to account for the strengthening or weak-
ening of an acid, is due to the amount of puU exerted upon electrons
by various atoms, it is evident that the amount of this pull cannot alone
determine the displacement of electrons with respect to the positive
kernels. for we must also consider the rigidity of the electronic stntc-
ture. In a molecule. or in a part of a molecule. where such rigidity
is small, or in other words where the electrons may he said to be
mobile, a small force- wn.l produce a reJa6vel_v Jarge displacement and
therefore largely influence their dissociation. Therefore we need not
he surprised. when we compare two different hydroxides. to find one
of them at the same time a stronger acid and a stronr;er base than the
other.
Although this is a subject whjch is stilJ ohscure. we cannot doubt
that there are molecules in which the framework of electrons is very
rigid. and that there are other molecules that are held hy less nowerful
constraints. 1\1"oreover we can state in general terms 'the criteria bv
which we may ascertain whethf'r or not the electrons are mobile or
are ri.ttidly held. Tn any molecule. or in any part of a molecule. which
has those properties inrlkativf' of what we call unsaturation. the electron
structure is in a mobile rondition.
For example, a doub1e bond produces such a condition of unsatura-
tion. In an nrp'anic molecule anv nel!8tive nHika1. such as O. OR. Nfl,
and eN produces unsaturation: A nd this is true when two neeative
elements are bonded tog-ether. causing that state of strain unon which
we have nreviouslv commented. Heavv radicals. such as iodine and
triphenylmethyl. cause unsaturation. and this is nrohahlv due in part
to the greater disruptive force in a heavy molecule which is in thermal
rotation. The loosening of the hond in molecular iodine and in hexa~
phenvlethane is shown both by the thermal and by the electrolytic dis-
sod:ttion of these substances.
When we sav that any double bond increases the mobility of the
electron ~tructure, we do not mean merely that the electrons which
produce the double bond are mobile. but rather that all electrons in the
neie:hborhood are less rigidly held. If we consider that every con-
dition of unsaturation is caused by some departure from the normal
state of the stahle pair and octet. we may suppose that the system will
assume a condition in which the total unsaturation is reduced to a
minimum. It is probable that this condition is best satisfied. not by
leaving most of the structure in its normal state and concentrating
the unsatnration at one point. but rather by distributing the neces-
sarv distortion to some extent throughout the molecule.
-Let us give an example to illustrate the necessity of considering
this factor of electron mobility when we are discussing the strength
of electrolytes. Methyl alcohol is a weak hase. By replacing the

methyl hydrogens by more negative groups we should expect the alco~
hoI to become an even weaker base. But phenyl is a negative radical
and triphenylmethyl carbinol, (C,H,),COH, is a vastly stronger base
than methyl ale-ohot. Here we have an extreme in.<;tance of the loosen-
ing of chemical bonds. The phenyl group, not only because of its
large size and weight, but also because of its double bonds, has a large
effect in loosening neighboring bonds. Three phenyIs upon a single
carbon atom give the very extraordinary radical triphenylmethyl, which
is capable of free existence as an odd molecule. and which is never
tightly held to any other radical. In the carbinol the hydroxyl group
may be regarded as being held by a very loose hand.
FinaUy it must he pointed out that even with all these reservations.
the electrochemical effect of a given radical cannot be quite so simple
as we have imagined it. In discussing the effect of substituted chlorine
in the chain of an aliphatic acid we have seen that the influence of
the chlorine bet'omes steadily less as the number of carbon atoms
separating 1t from the carhoxyl group increases. The action appears
to be through the chain, hut this view is not always tenable, as is
shown clearly by the case of maleic and fumaric acids, namely,
HCCOOH HCCOOH
HOOC~H II
and
HCCOOH
Here the n~lmher of atoms separating the two carboxyl groups is the
same in both acids, and we might therefore expect that the dissoda~
tion constants would be very nearly the same. As a matter of fact
the first constant of the cis-acid lS ten times that of the trans-acid,
showing that the two groups have a much greater influence on one
another in the former than in the latter. This influence is brought out
still more remarkably by the second dissociation constant. The ratio
of the first constant to the second is 4S in the case of the trans-
acid, but in the case of the cis-acid this ratio is 50,000. It seems evi-
dent that there is an opportunity for spatial approach of the two
carboxyls in the one case which is absent in the other, and that this
approach enormously heightens the mutual influence of the two groups.
The Rule of Crum Brown and Gibson.

We must not leave this subject without giving some attention to the
explanations which have been offered for the very remarkable phe-
nomena observed in connection with substitution in the benzene ring.
If one of the hydrogens of benzene is replaced by a radical of a cer-
tain class comprising such groups as Cl, OH, CH., a second substituent
enters the ring chiefly in the ortho and para positions. These radicals
are therefore known as ortho-para-orienting groups. On the other
hand, there is another class of substituents whieh cause the next sub-
stitution to Occur in the meta position. This class comprises such
radicals as NO., CN and COOH, which are known as meta-orientilll!
REMNANTS OF THE ELECTROCHEMICAL THEORY 145
groups. In one of the latest attempts to explain this phenomenon.

Stieglitz (1922), in terms of the mo~lern dualistic theory, assumed
that an ortho-para-orienting group makes the carbon, to which it
is attached, altogether positive. He then makes the ortho and para
carbons quadrinegative and the meta carbons bipositive. He thus
assumes the enormous difference of six units of charge between meta
carbons on the onc hand and ortho or para carbotls on the other hand.
Such an extreme electrical polarization, which far exceeds any that
we have been assuming, should produce startling effects upon the
strength of a carboxyl hydrogen. Indeed from all that we have seen
in this chapter we should expect even a relatively small difference
between the charge on an ortho carbon and that on a meta carbon to
'affect very greatly the dissociation of hydrogen from the carboxyl
attached to the atoms in these positions. In the nitro- and chlorhenzoic
acids we have an opportunity of testing these deductions, Chlorine
orients in ortho and para positions, The nitro group is the type of
meta orienting groups. I f Stieglitz' theory were correct we should
expect the chlorbenzoic aciel tn be a very strong acid in the meta ar-
rangement, and very weak in the ortho and para, On the other hand we
should expect ortho-nitrohenzoic acid to be very strong and meta-nitro-
benzoic acid to be very weak. The fact, however, i~ that both nitro- and
chlorbenzoic acid behave almost exactlv alike, The dissociation con-
stants for nitrohenzoic acid are: ortho; 6 x IO~a; meta, 3 x 10-4. For
chlorhenzoic acid: ortho, 1.3 x 10- 3 ; meta, 1.6 x 10-4.
Whether the negative substituent in benzoic acid is an ortho or
meta orienting group, its effect on the strength of the acid is entirely
analogous to the effect produced by negative groups in aliphatic acids.
The effect is large when the negative substituent is on the atom next
to the one which has the carboxyl; the effect is much smaller when
the substituent is removed by one more carbon atom in the meta ar~
rangement; and usually the para form is the weakest of all, although
in some cases it proves to be slightly stronger than the meta, which
might perhaps be considered evidence that there is a very slight alterna-
tion of charge in the benzene carbons.
It might be argued that, in both of the acids that we have been
discussing, it is the carboxyl group which determines the electronic ar-
rangement, but this argument can be met immediately by considering
benzene in which two hydrogens have heen replaced by carboxyls,
giving the phthalic acids. Here again we find just snch a behavior
as would be predicted from our knowledge of the dibasic aliphatic
acids. The first dissociation constant of the ortho acid is about five
times that of the meta acid, while the ratio of the first to the second
dissociation constants is 10 for the meta acid arid nearly 1000 for the
ortho acid, (In this respect ortho-phthalic acid resembles oxalic acid;
in the fonner, it is true, there are two more carbons between the
carboxyl groups, but they are atoms in which the electron mobility
is high.)
. An entirely different kind of alternation in the benzene ring has
been assumed by Fliirscheim (1902, 19(5), who considers that it
is an alternation, not of electric charge, but of Hresidual affinity." This

residual affinity corresponds to what we have called a state of un~
saturation. There is much to be said in favor of this hypothesis,
for indeed we see evidence not only of alternation within the ring,
as shown by the fact that an ortho-orienting group is also a para-
orienting group, but also in a chain directly attached to the benzene
ring. Thus we see that the OR group, with oxygen attached
directly to a benzene carbon, is ortho-orienting. The carboxyl radical,
in which the oxygen is attached to a carbon atom once removed, is
meta-orienting, while the radical CH 2COOH, in which the oxygen is at-
tached to a carbon twice removed, is again ortho-orienting.
This whole problem is one of much difficulty. It may be re-
marked, however, that we are dealing here with a phenomenon which
merely concenlS rates of reaction. An ortho-orienting group increases
the rate of substitution in the ortho position. There is no evidence
at present that the compound produced is any more stable than one
which would be produced by the corresponding substitution in the
meta position.
Volhile we may certainly conclude that electric polarization is small
in benzene and hs derlvatlves, nevertheless there is one point that must
not be overlooked. All of the methods we have used to determine the
degree of polarity of substances,-including the study of the disso-
ciation of electrolytes,-give information regarding the average state
of the molecules. However, when a substance enters into a reaction,
it may be only molecules in a quite exceptional state that take part.
At the recent conference of the Faraday Society (July, 1923) all of
those who participated seemed agreed that the average organic mole-
cule is very little polarized, but there were some who believed that
polarization and indeed ionization precede every reaction. This seems
too extreme a view; even when the breaking of a bond is the first
step in a reaction, the electron pair may be equally divided (e. g.
hexaphenyl ethane in benzene). Yet, especially in the more polar
solvents, a reacting atom may often get complete possession of a bond-
ing pair, thus causing some reaction. The momentary surge of elec-
trons that results in such polar breaking of the bond is apparently
favored in a molecule in which the successive atoms may readily
acquire a large and alternating polarity. This seems to me the basis
of the important theory of "induced alternating polarity" of Lapworth
and Robinson.
Resume.
We have seen in this chapter what a pitiful residuum is left to 11.
of the once powerful electrochemical theory. While electrostatic forces
evidently play an important part in processes of ionization, and very
likely also in numerous reactions which verge upon the ionic type,
such forces are responsible neither for the fundamental arrangement
of electrons within the molecule nor for the bonds which hold the
atoms together.
Chapter XIII.
The Source of Chemical Affinity; a Magneto-
chemical Theory.
I believe that enough has been said to show the incompetency of
simple electrostatic forces to account for the essential characteristics
of chemical combination. If a final argument be needed, let us com-
pare the atom of argon, with a positive nuclear charge of 18 and
its 18 electrons, with the atum of potassium which has a positive
nuclear charge of '9 and '9 electrons. It might be expected that
the removal of an electron would be about as easy from the one of
these atoms as from the other. Instead we find that it takes but
4 volts to ionize a potassium atom, while it requires IS volts to ionize
the argon atom, An en;n greater disparity is found between helium,
which has a nuclear charge of +2 with 2 electrons, and lithium, which
has a nuclear charge of + 3 with 3 electrons. The ionizing potential
of lithium is S volts, that of helium is 2S volts.
In our previous chapters it has occasionally been hinted that in
place of the electric it is the magnetic properties of the atom and the
mo1ecule which determine their essential structure. In the present
chapter we shall give free rein to this idea. 1
In Bohr's theory of the orbital electron each electronic orbit con-
stitut~s an elementary magnet or magneton. In the case of the simple
hydrogen atom the electron in its first orbit, or lowest energy level,
has the smallest magnetic moment; in the second circular orbit it
has twice that moment; in the third it has three times that moment,
and so on. It has been suggested that one of the quantum conditions
of atomic structure is that in a complex atom any magnetic moment is
either equal to that of the hydrogen atom in its most stable state or
an integral mUltiple of that value. In other words, if the hydrogen
atom at its lowest energy level is considered to have the unit of
magnetic moment, the moment of any other atom may be expressed
as an integer or zero. ,
If we consider two electrons in an atom, each of which has a
unit magnetic moment, the two together may give a magnetic moment
of 2 or 0, according to whether the two elementary magnets are so
oriented as to amplify or to nullify the magnetic effect. In his
remarkable treatise on atomic structure (1922) Sommerfeld discusses
the determination of atomic magnetic moments from spectroscopic data.
I have recently had the pleasure of a personal discussion on this sub-
ject with Professor Sommerfeld, and apparently it is safe to assert
1 Ramsay (19I6) suggested certain magnetic molecular models.
147
that an atom which possesses an odd number of electrons always has

a magnetic moment, while atoms with an even number of electrons
usually have no magnetic moment.
Here therefore we have very direct evidence that the pairing of
electrons, which I have regarded as the most fundamental phenomenon
in all chemistry, is some sort of conjugation of two magnetons, of
such character as to eliminate mutually their magnetic moment. This
conjugation does not always occur. In the elements of variable atomic
kernel, such as iron, we have already seen that both chemical and
magnetic properties indicate a lack of conjugation of the elementary
magnets, and in such atoms Professor Sommerfeld finds that the
spectroscopic evidence indicates the presence of several units of mag-
netic moment. There are also some molecules, like that of elementary
oxygen, which indicate a failure of the magnetons to couple, and we
shall see that this is probably characteristic of the double bond.
Nevertheless, we may state as the first law of chemical affinity
that electrons in an atom or a molecule tend to pair with one an-
other in such manner as to eliminate magnetic moment. The odd
molecule tlwrefore represents the highest degree of chemical un-
saturation. Even the odd molecules which have been isolated in the
free state show as a class the properties which we should expect of
such substances. If we could work under ordinary conditions with
such substances as free methyl, monatomic hydrogen and monatomic
chlorine, we should find these substances to have a reactivity such as
no existjng substances possess.
The a('t11al uiSlX)sition of the extra electron jn an odd molecule
will not be an easy matter to ascertain~ but it presumably seeks that
portion of the molecule which possesses the highest degree of un-
saturation, and there orients itself in such manner as to reduce the
total unsaturatiQu to a minimum. So also an odd molecule attaches
itself readily to other molecules, even to those which we regard as
least unsaturated. So, for example, triphenylmethyl forms a compound
whh hexane.
The odd molecule of sodium dissolves in ammonia or an amine to
form a compound from which sodium ion may dissociate, leaving the
odd electron attached to the solvent, as is shown by the investigations
of Gibson and Argo (1918) on the absorption of light by these sub-
stances. Monatomic hydrogen attaches itself to the hydrogen molecule
H, to produce the odd molecule H,. It has been suggested that this
molecule of ,"active hydrogen" is to be represented by a symmetrical ring
structure, but jt seems far more Jikely that it is a loose combination
between Hand H, entirely analogous to the compound between tri-
phenylmethyl.and hexane.
The fact that some odd molecules actually exist in the free state
shows that there are conditions under which a molecule which has
no magnetic moment may break into '"two molecules, each of which
has a magnetic moment, as hexaphenylethane dissociates In form Iti-
pheny1methyl, and as iodine at high temperatures assumes the mon-
atomic state. This dissociation is evidently the result of thermal
THE SOURCE OF CHEMICAL AFFINITY 149
agitation, but there are other more obscure factors involved in that
loosening of the chemical bond which permits such dissociation. \Ve
should expect that heavy radicals would be subject to a greater cen~
trifugal force and therefore dissociate more readily, and we find that 12
dissociates more readily than C12 • and <PsCCtpa dissociates more readily
than H,CCH,. But, as has been especially pointed out to me by Pro-
fessor Branch, the loosening of the bond in hexaphenylethane is not due
alone to the weight of the radical, but also to the unsaturated character
of the phenyl group, that we discussed in the last chapter. If in
hexaphenylmethane we should replace the benzene rings by hexa-
methylene rings, the radkals would be slightly heavier than before butJ
there would be no appreciable dissociation.

\Vhen two odd molecules, each having a magnetic moment, unite
by the coupling of their odd electrons to form a system in which
the magnetic moment is eliminated, the energy of the separate magnetic
fields may be considered as set free, manifesting itself as the heat of
combination. In general ' . . e may suppose the state of minimum energy
to be a state of the utmost mutual neutralization of magnetic fields,
except in so far as electrostatic forces and possibly some other factors
of e\'en less significance intervene.
\N'hen all the electrons in a molecule are paired, even if we assume
that the magnetic moment of a bonding pair and of every other
electron pair is exactly zero, it would be neither necessary nor
desirable to assume that the magnetic fields of the paired electrons
completely neutralize one another. 1n other words, we may assume
the complete disappearance of magnetic moment (although this has
not yet been definitely proved), and at the same time we may assume a
residual magnetic field emanating from the electron pair. This stray
field we shall now consider to be equivalent to what has been called
residual affinity, and to be responsible for that condition which we have
spoken of as a condition of unsaturation. This residual affinity is
therefore of a very different kind from that which would be due to
electric polarization, where it would be necessary to assume two kinds
of field, one emanating from a poshive and one from a negatlve charge.
The several pairs of electrons in an atom may arrange themselves
in such an orientation as to neutralize still further their residual fields.
The second fundamental principle of chemical affinity may therefore
be stated as follows: Every atom, except that of hydrogen or helium,
has the smallest external magnetic field, and is therefore in a con-
dition of maximum stability, when it possesses in its outer shell four
pairs of electrons situated at the corners of a regular tetrahedron. The
condition of saturation accompanying such an arrangement is best ex-
emplified by the atoms of the argon type. The noble gases in their
diamagnetic behavior and in their chemical inertness show a nearer
approach to complete saturation, or lack of residual affinity, than any
other known substances.
Whenever such a group of eight (or group of two in the case of
hydrogen) is effected by the sharing of electron pairs, there is a
less complete neutralization of the residual magnetic fields, and we
ISO VALENCE AND THE STRUCTURE OF ATOMS AND MOLECULES
may say that even a substance which contains only single bonds is not
a fully saturated substance. Such single bonds as in hydrogen-hydro-
gen, hydrogen-carbon. and carbon~carbon seem to possess the smallest
residual magnetic fields. When one of the bonded atoms is that of a
negative element the residual field is greater, and becomes specially
pronounced when two such elements are bonded together.
Whenever the symmetrical and stable structure is in any way dis~
totted it leads to an increase in the residual magnetic field, and a dis~
tortion produced in one portion of a molecule will ordinarily cause
some distortion in neighboring atoms in such manner that the total
resultant unsaturation is reduced to a minimum. Such a distortion
we find for example in the unsaturated substances with rings of three
and four carbon atoms. All such distortions of the molecule, with
the accompanying increase in residual magnetism, may be considered to
be due to the change in the relative position or orientation of the sev-
eral pairs, or to the partial "opening up" of a single pair, and probably
either of these effects ordinarily accompanies the other. Such a dis-
placement of the electrons from their most stable positions may be
taken as equivalent to what we have called the loosening of bonds and
the increase in electron mobility.
During the whole course of my jnvestigations relating to the struc-
ture of molecules a very interesting question ha~ constantly recurred
to which I feel unable to give a definite answer. In our discussion
of the electrochemical properties of substances we have assumed that
if one pair of electrons, were drawn away from an atom in a certain
direction by an element striving for sole possession of the bonding pair,
then the remaining pairs of that and neighboring atoms would be
drawn in the same direction by electrostatic forces. This, however,
would place the octet as a whole in an unsymmetrical position with
respect to the atomic kernel, and the question is whether we may not
occasionally meet with just the opposite phenomenon in which the
drawing out of one or more pairs from the atomic center might lead
to a withdrawal of the remaining pairs, in such manner as to result in a
symmetrical tetrahedron about the kernel. If three of the electron
pairs are drawn away from an atom through union with electronegative
atoms, the two possibilities would be represented by the following
scheme:
: X: :X:
Thus if the three hydrogens of methyl alcohol are replaced by
phenyl groups these will draw three pairs of electrons away from
the carbon atQ.m) and if we were to assume that the remaining pair,
which is the bond to the hydrogen group, were forced away from
the carbon atom at the same time, it m!ght in this way be possible to
account in part for the notably basic properties. of triphenylmethyl
carbinol which we recently considered~ This question is closely con-
nected with the possibility of alternation of properties in a chain of
carbon atoms, which we discusseol in the preceding chapter.
THE SOURCE OF CHEMICAL AFFINITY lSI'·
It may be that under certain circumstances two atoms which share

a condition of unsaturation may reach their condition of greatest
stability, not when the unsaturation is distributed between the two
atoms, but rather when one of the atoms approaches its normal state
and the other bears the full brunt of the unsaturation. Stieglitz·has
explained certain phenomena by stating that an atom tends to be~
come wholly positive or wholly negative. While we have seen that
the assumptions underlying such a statement are untenable, might
we not assume that of two adjoining atoms one tends to be wholly
saturated while the other carries the whole or the greater part of the
residual affinity? While there .are a number of facts which sug-
gest this possibility I have sought in vain for any definite evidence that
J
such a phenomenon exists. \\o'e might consider the symmetrical and

unsymmetrical forms of di-chlor ethane. The heat of formation as
far as we can ascertain from existing data is about the same for
both substances. As to their magnetic properties Pascal has shown
that the symmetrical form is the more diamagnetic of the two, which
seems to indicate that the unsaturation is least when it is shared
between the two carbon atoms.
With the exception of the odd molecules, the molecules of organic
substances which we must regard as most unsaturated are those to
which we ascribe multiple bonds, and espec~ally double bonds. Never-
theless a molecule of the ethylene type is not so unsaturated as we
should expect it to be if we had to assume the two carbons to be
united by a single bond, for tben we sbould be obliged to assume
either two odd electrons or a highly unsymmetrical and electrically
polarized molecUle with eight electrons on one carbon and six on the
other. These facts, together with the existence of cis and trans isomers,
lead us to assume the double bond. However, the exact physical sig·
nificance of the double bond remains somewhat mysterious. It is
certainly not the equivalent of two single bonds, but rather indicates
a state intermediate between this and the condition which would exist
if there were but a single bond and a consequent deficiency of electrons
to make up the two carbon octets.
Both the chemical and the magnetic properties of substances with
double bonds show a high degree of uns.turation. The extremely valu-
able investigations of Pascal show that the diamagnetic susceptibility
is far less than would be calculated from the addition of atomic sus-
ceptibilities whenever we have a double bond between carbon and carbon,
carbon and nitrogen, nitrogen and nitrogen, or carbon and oxygen.
The most striking case is furnished by the double bond between oxygen
and oxygen in the molecule 0.. Molecular oxygen is a remarkably
paramagnetic substance.
While we must recognize the extraordinary parallelism between
diamagnetism and chemical saturation, we cannot hope to give a c0m-
plete interpretation of the latter in terms of the former until we
possess a more adequate physical theory of diamagnetism. The simplest
theory of this phenomenon would make the diamagnetic susceptibility
of a molecule proportional to the total number of electrons in all the
E AND THE STRUCTURE OF ATOMS AND MOLECULES
severa} atoms. This, however, is certainly not the case,

; found that the diamagnetism of an atom in<;-reases roughly
mic number. Thus according to Pascal, the accompanying
he atomic susceptiblHties of a number of elements. It is
TABLE
ATOMIC SUSCEPTIBILITIES (X 10. 1),
Hydrogen .... -30·5 Phosphorus ..... - 274

Carhon -62.5 Sulfur ........ ,'.... -156
Nitrogen ............. , .. , -58 Chlorine ................. - 209.5
Uxygen ................. . -48 Bromine ................. - 319
l<luorinc ..... '.' ., ......•. -65.5 Iodine ,.................. - 465
evident that there is no simple relation between the susceptibilities and

the numher of electrons.
The simplest theory of paramagnetism would be that this phe-
nomenon is non-existent except when the molecule has one or more
units of magnetic moment. The susceptibility measured by the Drdj~
nary methods would be the algebraic sum of the susceptibilities due
to the diamagnetic and the paramagnetic behavior of the molecule.
The idea that the magnetic moment of an atom or molecule can
change only by integral steps is no essential part of our magneto-
chemical theory, As far as concerns the facts of chemistry and many
of the facts of magnetism, it would be simpler to assume that any
condition which causes unsaturation by opening up the condensed mag-
netic system, of electron pair or octet, produces some increase in para-
magnetism, thus diminishing the diamagnetism which would be attained
in the absence of any magnetic moment.
On the other hand, the assumption that the magnetic moment can
occur only in discrete units would lead us to believe that diamagnetism
is diminished in some unknown way whenever there is a distortion
of the normal magnetic system, and that as this distortion increases a
new phenomenon may suddenJy appear, when tbe distortion reaches the
point of setting up a magnetic moment.
These are questions which we ought soon to be able to decide
by experiments of the type of those of Stern and Gerlach, but 011r
present information is inadequate. It is to be noted that any experi-
ment depending upon the ohservation of the average properties of a
large assemblage of molecules might indicate a gradual increase or
decrease in total magnetic moment without necessarily discrediting the
theory of discontinuity in magnetic moments, for there might he a
tautomeric equilibrium between molecules having a magnetic ffiO!llI!11!.
and others having none. The experiments might merely indicate a
change in the relative amounts of the 1\-,"0 tautomers.
Undoubtedly the molecule of oxygen possesses a magnetic moment,
and it seems probable that this is also true in other types of double
bond. We could assume the double bond of ethylene broken in ~
manner as to give au odd electron to each carbon atom, so that eaCh
atom would have a magnetic moment. But this is not the only way in
which such moments could he set up. We have spoken several times
of the possibility that the electrons in the atom may under certain
circumstances be removed from the primary to a secondary valence
shell. If we were to employ the analogy o.f the hydrogen atbm as
portrayed by Bohr, we might consider that a single electron in the
first valence shell would have one unit of magnetic moment, but in
the second shell might have two units. If one .electron of each of
these types were then to conjugate we might ha"'e a pair of electrons
differing from an ordinary electron pair in that it possessed a resultant
magnetic moment of one unit.
However, any answer to these questions must at present be RpeCU-
lative and need not greatly concern us. for it is sufficient to our present
purpose to recognize that anything which diminishes the diama{.,rnetism
of a molecule increases the residual affinity or the condition of
unsaturation.
, As compared with<the double bond, a triple bond produces very little
diminuf'i011 £n diamagnetism., This is shown not only by the investiga-
tions of Pascal on compounds containing triple bonds between carbon
and carhon, and carbon and nitrogen, btit also by the diamagetism of
the nitrogen molecule itself. Substances with triple bonds represent
relatively saturated structures.
Conjugation.
In the course of this book we have been using the word conjuga-
tion in what are apparently two different senses, We have spoken
of that conjugation of two double bonus which diminishes the un-
saturation of a molecule, and we -have spoken of two odd electrons,
each having a magnetic moment, conjugating to produce a couple which
is largely self-coqtained tnagnetically~ and possesses little residual
magnetic field. We are now in a position to assert that these two
meariings are identical, and may say that every process which leads
to a partial neutralization of molecular magnetic fields is a process
of conjugation. It probably would not be far from the truth to
state that nearly every chemical process occurs in such manner as
to increase the net amount of conjugation.
When two unpaired electrot:ls combine to form an electron pair
we have an extreme type of conjugation. When four of these pairs
arrange themselves at the comers of a regular tetrahedron, in order
to neutralize further their residual magnetic fields, this also is con-
jugation. When any distortion of such a symmetrical structure is re-
Heved. there is again conjugation. When a molecule rearranges SO
that the loose bond between two negative elements is broken and
firmer bonds are produced, there is marked conjugation. Also in
the conventional sense of the organic chemist we may say that when
two pairs of doobly bonded carbon atoms are separated by a single
bond the molecule is able to rearrange itself, to diminish the strong
residual magnetic fields and form a conjugated system.
There is every indication that the effect which we call residual

affinity or stray magnetic field manifests itself but a short distance from
the seat of unsaturation. We may therefore conc1uge that when two
residual fields, which in our method of writing chemical formulre
appear to be widely separated, conjugate with one another, there is 'a
spatial arrangement of the molecule which brings the two centers of
un saturation close together. The possibility of such approach will
often determine whether or not conjugation occurs, and we see in
maleic and fumaric acids, which we discussed in the last chapter, two
suhstances which would doubtless behave alike were it not for the
opportunity of conjugation afforded in the cis form.
Benzene is the typical highly conjugated system. It not only shows
little chemical reactivity, but it also shows, according to Pascal, hardly
any of the diminution of diamagnetism which would be expected of
a system containing double bonds. From its chemical and magnetic
properties, it seems to me that we can state with certainty that the
unmodified Kekule formula is not the true formula for benzene,
which undoubtedly has a more compact molecule than that formula
would imply.
We have seen, in Figure 2-+, the electron arrangement which Hug~
gins has prolxlsed for the benzene molecule. \Vhether this model is
COffect we cannot at present say, but it seems certain that ehher this
or some similar structure will prove to express best the varjous prop-
erties of the benzene ring. Perhaps the true formula will be found
to be one which is in some respects intermediate between this model
and the one of KekuJe. The symmetry of the benzene structure un-
doubtedly permits a degree of conjugation which would not other-
wise appear. Indeed we find that dihydro- and tetrahydrobenzene are
far more unsaturated, chemically and magnetically, than benzene itself.
One of the most interesting types of conjugation is that which
occurs in the carboxyl radical. The ordinary formula given to an
organic add,
R
O=COH,
would indicate the existence of a typical double bond between carbon
and oxygen. On the other hand we have a large amount of evidence
that this is not a typical carbonyl uuion. Various physical methods
that have been applied to organic substances indicate this. Pascal
shows that the diamagnetism is much greater than would be predicted
for a carbonyl compound. Certain chemical facts further support
this view. We have seen that elements like silicon and sulfur rarely,
if ever, exhibit the double bond, but in the carboxyl radical we may
snbstitute silicon for carbon or sulfur for oxygen.
The conjugation apparently may occur in two ways: either two
carboxyl groups may conjugate with one another, or if this does
not occur, a single carboxyl group attains by itself a condition which
is far more saturated than can be represented by a formula with a
double bond. The conjugation between two carboxyl groups may
occur when both are in the same molecule, and the facts presented in
the preceding chapter indicate such a conjugation between the two
carboxyls of maleic acid. It may also occur when the two carboxyls
are in different molecules. A substance like acetic acid is dimolecular,
not only in the liquid state, but also to a very surprising degree in
the gaseous state. The union between the two molecnles must almost
certainly occur at the carboxyl groups, and these two groups are to
be regarded as held together by a conjugation of some sort.
A n
FIG. 26,~Conjugation oi the Carboxyl Group (Huggins),
As to conjugation of a single carboxyl, Huggins bas offered a

suggestion whkh is interesting, and which may rome very ncar to
representing the truth. This model, with slight alterations, is shown
in Figure 26B, while Figure 26A represents- the structure ordinarily
assumed. In both cases each hlack circle indicates a pair of electrons.
Tbe cbange consists in bending tbe OH group until the bydrogen is
equally placed between the two oxygens, and the whole radical is
symmetrical, so that there is no longer one doubly hound and one
singly bound oxygen. The known facts regarding the isomers of sub-
stances of this type strongly support such a theory of conjugation. If
we replace one oxygen in carboxyl by the NH group we might expect
two isomers, namely,
R R
HN=COH, H,NC=O.
Such isomers are entireJy unknown, nor would they be expected from
the formula of Figure 26B, for the acid hydrogen would belong equally
to the oxygen and tbe nitrogen. (The other hydrogen must ordinarily
remain attached to the nitrogen. for its transfer to the oxygen would
mean very large electric polarization.)
Conjugation in the carboxyl group seems to be closely paralleled in
certain tautomeric substances containing a chain of three carbon atoms,
in which the traditional method of writing organic formulre shows a
mobile bydrogen attached to one of the outer carbon atoms while tbe
remaining two are united by a double bond. However, in an extremely
convincing paper just presented by Tborpe and Ingold before the Inter-
national Union of Pure and Applied Chemistry (1923), it is shown
that in substances of this type, such as tbe glutaconic acids. the normal
structure of the molecule must be a symmetrical one in whicb tbe mobile
hydrogen is equally shared by the (l- and y- carbon atoms.
In all such cases of conjugation it is evident tbat we are dealing
with conditions which cannot be adequately represented either by

old-fashioned bond formulre or bv mere translation of the old formulre
into our present theory. by substituting electron pairs for bonds. In-
deed even in such models as those of Figure 26, we must remember
that the tetrahedra which are introduced to facilitate visualization of
the model have no real existence. A true picture of a molecule wo~ld
show only the positions and orientations of the atomic kernels and the
electrons... It js further to be borne in mind that no one model could
adequately represent the structure of a substance in which there is
tautomerism between various molecular structures.
Not only may double bonds conjugate with one another, but one
double bond may conjugate with any other source of residual affinity.
Such a conjugation with hydroxyl we have just discussed in con~
nection with the carboxyl group. An essentiaJJy similar conjugation
is assumed by Fliirscheim in his theory of substitution in the benzene
ring. There it is assumed that the residual affinity of the first subw
stituent conjugates with the adjacent double bond in such manner
as to alter the original conjugation of the benzene ring. That the latter
is in an extremely labile condition, is shown by the fact that the small
change in residual field produced by the mere exchange of a methyl
group for a hydrogen suffices to alter the state of conjugation.
Any bond between carbon and a halogen produces a large residual
magnetic field, as is well shown by Pascal's discovery of the lack
of additivity of the diamagnetism of organic substances containing
halogens. The possibility of conjugation between a carbon-chlorine
bond and a double bond is illustrated when the four hydrogens of
ethylene are substituted by chlorine. The compound Cl,C = CCI,
shows a far lower degree of chemical reactivity than is usually found
in molecules with double bonds. We have noticed several facts which
show that what we call a triple bond is more saturated than a double
bond. We are thus ahle to understand a peculiar form of conjuga-
tion which apparently consists in the conversion of a double bond into
a triple bond. The conversion of diazo-compounds into diazonium
compounds seems to belong to this category. The probable course of
this reaction is indicated by the following scheme:
:O:H
<p:N: :N: = [<p:N:: :N:T + [:q:HJ
In whatever manner conjugation occurs, we must realize that if a
system possesses two or more s~urces of unsaturation, the total
unsaturation may be the sum of that due. to the separate sources, but
wHl be less whfnever there is any opportunity for such spatUd re-
arrangement a'lld reorientatio'll as will serve to neutralize the residual
magnEtic /kIds. Indeed it is the process of such rearrangement that
we know as conjugation.
Chapter XIV.
The Discontinuity of Physico-Chemical Processes.
Attempts to apply quantum theory to chemical reactions have been
largely limited to a study of reactions produced by light and of the
light set free by chemical reactions. It has long been known that
numerous photochemical reactions which occur in the presence of blue
or violet light occur to a far more limited extent or not at all in
red light. So also many reactions which are not produced by visible
light take place upon exposure to ultraviolet radiation.
If one or more of the reacting molecules must receive a certain
quantity of energy before it can react, then we should expect from
quantum theory that no exposure to radiation, no matter how pro-
tracted, would cause the reaction to occur, unless the frequency of
the radiation were high enough to make hv as large as the energy re-
quired by the reacting molecule. It is true no case is known of a
typical photochemical reaction which occurs rapidly with light of a
given frequency and which does not occur at all with light of a little
lower frequency. Nor perhaps is this ordinarily to be expected, ~ince
different molecules, owing to thermal agitation, would require some-
what different amounts of energy to reach a condition in which they
would react.
The interesting idea has been suggested by W. C. Mce. Lewis
(1916) and by Perrin (1919J that all chemical reactions are photo-
chemical in character. They assume that a moIecule does not r~ct
until it becomes activated by radiant energy of a certain minimum
frequency. This light may come from outside the system, as in the
typical photochemical process, or it may exist in the interior of the re-
acting system as thermal radiation. The relative content of high
frequency radiation, in general thermal radiation. increases very rapidly
with the temperature, and it has been shown by Lewis that his as-
sumption leads quantitatively to an equation which agrees with the
equation that Arrhenius obtained for the change in reaction velocity
with the temperature.
The assumption that every simple chemical reaction is accompanied
by the absorption of light of one frequency and the emission of light
of another frequency enabled Perrin to give a beautiful explanation
of tbe phenomena of photo- and thermo-luminescence. Nevertheless in
spite of the great value of this work in pointing out the influence of
radjation in chemical phenomena, we cannot adopt the main conten-
tion that reactions are due solely to the influence of light and not
at all to the molecular bombardment due to tbermal motion.
157
This is perhaps sufficiently shown hy a consideration of those simple

phenomena which we may regard as the prototypes of all chemical
processes, namely. the resonance and ionization of gases. These phe-
nomena are known to be caused either by radiation, or by "moving
electrons, or by alpha particles, and they presumably can be caused
by any other kind of molecular bombardment.
The Discontinuity of Chemical Processes.

If we were to consider anyone philosophic idea as the leading
principle in the scientific thought of the last two generations it would
be the belief in the continuity of nature. The concept of energy and
of its flow through material systems and through free spsce, the de-
I;'
..1 Ir~/ V- V
~
~
[......./
.~
" r-
[]
Kinetic Energy of Actlvatitlg EJectmlls --+
FIG. 27.-Synthesis of Ammonia by an Electron Stream.
velopment by Maxwell of the idea of the electric and magnetic fields,

with the discovery of relativity by Einstein, all contributed toward
a marvellously simple picture of the universe, based upon the under-
lying theory of extension in a continuum.
Chemists and physicists, when they have plotted their experimental
data and have obtained curves with breaks corresponding to no obvious
discontinuity in the system studied, bave attributed such breaks to
experimental error. In nearly all cases this explanation has been justi-
fied, but there are instances, and these are now becoming more numer-
ous, in which such broken turves are found to possess real significance
even in systems which seem to possess no apparent source of dis-
continuity. Professor Olson and Dr. Storcb have kindly permitted
me to reproduce in Figure 27 some of !:be curves which they are about
to publish, giving the results of their study of the union of, nitrogen
and hydrogen, to form ammonia, in the presence of a stream of electrons.
The ordinate gives the rate of the reaction, and the abscissa the kinetic
THE DISCONTINUITY OF PHYSICO·CHEMICAL PROCESSES, 159
energy of a constant number of moving electrons. As the latter

changes graduaUy the former changes in a manner whi<:h indicates at
first sight a series of very inaccurate experiments. The successive
steps in the curve, however, are reaJ and have been reproduced at will.
Such a result is not essentially different from that which is observed
in the resonance and ionization of a simple gas. The fact that
electrons moving with a velocity less than a certain critical velocity,
or that radiant energy with a frequency less than a certain critical
frequency, produce no change in a molecule, while a slightly higher
velocity or a slightly higher frequency of light produces a profound
change in the molecule, must be regarded as one of the cardinal facts
in all chemistry.
As far as we can see, a hydrogen atom which is in the first
or lowest energy state can undergo no change whatsoever unless
it can acquire a quantity of energy sufficient to raise its electron from
the first to one of the other fixed energy levels, or to remove it altogether
from the atom. May we not therefore conclude that a similar state·
ment is true of any molecule, and that every reaction,'no matter how
complicated, takes place in one or more definite steps? Radical as
such a conception is, we may show that many of the chemical ideas
which have been developed and stated in terms of the continuou~
theory may be readily translated into the language of the discontinuous
theory.
We have often had occasion to speak of tight or loose honds, or
of electron pairs which are held in position by large or small constraints.
In the older theory this would imply a mechanical system in which
gradual displacement from an equilibrium condition would build up a
restoring force. The ratio of the restoring force to the amount of
displacement would measure the magnitude of the constraint, so that
a tight bond would be one in which a slight displacement would cause
a large force of restitution. We could express this same idea in an-
other way by calling an electron pair mobile when held by small
constraints.
In the theory of discontinuity there would be no such thing as a
gradual displacement within a molecule. The molecule must be in
one or another of a series of states with finite differences in energy
and in other properties. However, if we consider once more the case
of the simple hydrogen atom, we note that a large energy change is
required to remove the electron from the first to the second energy
level, but if the electron were in the twentieth level it would take
but a small amount of energy to raise it to the twenty-first, or indeed
to any of the infinite number .of higher levels, or to drive it completely
away from the' hydrogen nudeus. If we choose to employ the older
phraseology we may say that the electron is held tightly in the lower
energy levels and loosely in the higher levels.
In general we may conclude that when we speak of a loose bond-
ing pair, or when we say that a molecule or a certain portion of a
molecule is in a mobile condition. we mean that a state exists in which
small increments of energy suffice to cause a transfer to other neigh-
boring states. The loosening and the eventual rupture of the bond
jn the molecule of iodine or of hexaphenylethane may be considered
analogous to the resonance and eventual ionization of the atom of
hydrogen.
But while there arc some molecules like the hydrogen atom, or the
argon molecule, or the methane molecule, which in their most stable
states are far removed in energy content from the next possible state,
there are others in which even the most stable state is not very differ w
ent from other possible states, so that even a mild excitation such a5
can be produced by light of low frequency, or by a moderate tempera-
ture, gives rise to a number of different molecular states. The
molecules helonging to these several states could be caned tautomeric.
Let us consider a chemical reaction which involves merely the re-
arrangement of a molecule, such as the conversion of one optical isomer
into the other. Such racemization will occur more rapidly the looser
the structure is. In other words, the reaction is rapid when a slight
molecular impa,ct, or thermal radiation of low frequency, both of which
are associated with a low temperature, suffice to carry the molecule
through the several energy levels to the second stable state.
Probably the majority of chemical reactions, even those which
eventually lead to the evolution of large amounts of energy, require
some initial excitation of the molecule, and it undouhtecl1y frequently
happens that a system mus.t be 1i fted far above its original energy
level before it call fall illto a ,till lower energy level. A catalyst may
be regarded as any substance which, by forming a complex with one
of the reacting substances, or by any other means, introduces other
series of energy levels which permit a short-circuit of the normal
path of the reaction, in such manner as to diminish the amount of
excitation required.
Color.
The older theory furnished an extremely happy explanation of the
origin of color and its relation to the chemical properties of substances,
as r attempted to show in my paper on the U Atom and the Molecule,'"
According to that view, an electron in a position of constraint possesses,
like any other elastic system, a natural period of vibration,-this fre-
quency being proportional to the magnitUde of the constraint. Light
which has this same period of vibration is capable of imparting energy
to the electron resonator, and thus the electron absorbs light in that
part of the spectrum which corresponds to its own natural frequency.
In the majority of substances the electrons are held so tightly that
their natural frequency falls in the ultraviolet region, and these are
therefore incapable of absorbing visible light. In other words, they
are colorless. BMt under conditions which loosen the constraints within
the molecule the frequency of the electrons is lowered until it cor-
responds to a visible frequency. The substance thus absorbing some
part of the visible spectrum, and letting through other parts, is said
to be colored.
Since those changes in the condition of a molecule, such as the sub-
THE DISCONTINUITY OF PHYSICO.CHEMICAL PROCESSES .6.
stitution of one radical for another, which are known from chemiCal
evidence to loosen the structure of the molecule, and to render it more
unsaturated and more reactive, prove to be the same kind of changes
that convert a colorless into a colored substance, or which convert a
substance with absorption in the violet into a substance with absorp-
tion in the rcd, this explanation of color seemed to be eminently sat-
isfactory. Nevertheless the idea of a vibrating electron seems to be
one which is· not only alien to, but essentially incompatible with, the
spirit of quantum theory.
Fortunately the new theory of the discontinuity of chemical states
furnishes a very similar and equally satisfactory explanation of color.
When a molecule is in a state such that it may be changed to an-
other state by a quantity of energy which, when divided by the Planck
constant h, gives a mtmh~r corresJXlnding to the frequency of visible
light, the substance is colored. Any process by which a colorless sub-
stance is converted into a colored substance may be regarded as one
in which the molecule may be changed from one energy state to another
state of only sligHtly different energy. In other words. with a new
interpretation of our terms, we may still say that a colorless sub-
stance is converted into a colored substance by a loosening of the
electronic structure.
If we consider the halogen group, we may assume that the color of
the molecules of these elements is due to the looseness of the bond,
and the pair of electrons which constitutes the bond may be considered
as passing from one energy position to another during the absorption
of light. The bond is weakest in the case of iodine, and here we have
absorption of red light. As we pass through bromine and chlorine we
see evidence of a tightening of the bond. and in fluorine, where we
say that the bond is tightest, only the extreme violet end of the spectrum
is absorbed.
An colored substances are highly unsaturated and correspond to a
low state of conjugation. All known odd molecules, except nitric oxide,
absorb light in the visible region. :Most organic substances which
absorb light are those which contain double bonds. The benzene ring
itself is too highly conjugated to absorb visible light, but in the quinoid
form, which is far less conjugated, we find a configuration which almost
invariably causes color.
The great majority of compounds of the elements of variable kernel,
which appear in the transitJon regions of the long periods of the
Mendeleeff table, are colored. Here we have a case where even the
levels corresponding to inner and outer shells do not differ much
in energy. This is shown by the small amount of energy required
to effect the oxidation or reduction of such compounds. Thus a mild
oxidizing agent suffices to change a ferrous into a ferric salt, although
this involves the transfer of an electron from the iron kernel.
According to the theory of discontinuity, every absorption band must
consist of a series of absorption lines, except for the blurring effect
due to thermal agitation; and each of these lines must correspond to
the change of a molecule from one definite energy state to another,
I
~CE AND THE STRUCTURE OF ATOMS AND MOLECULES
\
\t that in many cases there must be a very large number of
\ energy states. Only in such a way can we account for
'omplex absorption spectrum of a substance like iodine.
__ ' case of the simple atom of hydrogen we are obliged to
..relIeve in the existence of an infinite number of energy levels, all
of which, except the few lowest members, correspond to nearly the
same energy. In a molecule containing a number of atoms, and many
electrons, the complexity of the system of energy levels is presumably
far greater. In a molecule which is highly unsatorated we may
assume that even the most stable state is one of many states which
arc very near to OTIe another in their energy content. ,But in the case
of a highly saturated or highly conjugated system we may assume
that the normal state of the molecule cannot be changed witbout a
large addition of energy.
Just as the absorption of visible light is an index of a very loose
electronic structure, so we should expect from the Einstein principle
that similar conclusions could be drawn from the reaction between
a molecule and a moving electron. The fact that iodine vapor absorbs
visible light would lead us to predict that a slow-moving electron would
be able to cause a change in the energy state of the iodine molecule.
Such a reaction might involve the attachment of the electron to the
iodine, forming a negative ion, or it might simply mean an inelastic
collision which would deprive the electron of a part of its kinetic
energy.
Both of these phenomena doubtless occur, and we are beginning to
get very valuable data concerning the movement of electrons in various
gases. All of these experiments indicate that moving electrons stick
to, or are retarded by, those molecules which also give chemical, optical,
and ,magnetic evidence of unsaturation. Thus by assuming that elec-
trons stick to molecules of a halogen, Gibson and Noyes (I922) have
offered a very plausible explanation of the disappearance of character~
istic metallic spectra in flames \1I,"hen free halogens are present. The
mobility of electrons has been shown by Wahlin (1922) to be greater
in ethane than in ethylene and greater in ethylene than in chlorine.
This is also the order in which we would place these substances with
respect to their degree of saturation. Likewise we should expect
the mobility of electrons to b~ far greater in a noble gas than in
any other type of gas, and this has been proved experimentally by
Townsend and Bailey (")22).
The Future of Quantum Theory.

In that old American institution, the circus, the end of the per-
formanee finds the majority of spectators satiated with thrills and
ready to return to more quiet pursuits. But there are always some
who not only remain in their seats but ~ make further payment to
witness the even more blood-curdling feats of the supplementary
performance.
Our own show is now over, and I trust that the majority of readers
THE DISCONTINUITY OF PHYSICO-CHEMICAL PROCESSES 163
who have had the' patience to reach this point will now leave the tent;
for what I am about to say is no longer chemistry, nor is it pbysics,
nor perhaps is it sense. But since we have been obliged here and
there to take cognizance of the entering wedge of scientific bolshevism,
which we can quantum theory, or the theory of discontinuity in nature,
1 cannot refrain from attempting to forecast some of the logical con~
sequences which must follow from the new facts that have been dis-
covered and the interpretation which they have been given. Such a
forecast must of necessity be of the crudest sort, and can hardly do
more than indicate the magnitude of the revolution in scientific thought
which probably must occur before physical science can once more be
a homogeneous whole, free from the most glaring inconsistencies and
contradictions.
Two quantitative methods have been available to scientists. One
consists in counting and the other consists in measuring. The former
has been the basis o{ the theory of numbers, the latter has led to
the development of geometry. The first of these sciences has been
the mere plaything of abstruse mathematicians; the second has become
the working tool of the scientist and the engineer. Geometry is based
on the theory of the continuum, and so also is the closely related science
of calculus. \'.Ie have been taught that an integration of the jn-
finitesimal elements of a continuum may be approximately replaced
by a summation of finite terms, but that the former method is exact
and absolute while the second gives but an approximation. Are we
not now going to be obliged to reverSe this decision and to recognize
that the branch of mathematics which will come nearest to meeting
the needs of science will he the theery of numbers, rather than a theory
of extension, and that measuring must be replaced by counting?
The mathematics of hydrodynamics is based on the theory of
the continuum. It is admirably suited to express the behavior of sub-
stances like water and air. Nevertheless, the method is entirely an
approximate one, for water and air are not continua but are com-
posed of discrete molecules. Hydrodynamics could not account for
such a phenomenon as the Brownian movement.
The methods of hydrodynamics were taken over into the field equa-
tions of electromagnetics. An electrostatic field, regarded as a con-
tinuum, is defined by the force exerted upon an infinitesimal test charge
placed within it. But an infinitesimal test charge is a concept which
we can no longer hold. The smallest charge is the charge upon a single
electron, and if we use the electron as ·a test charge to determine
the properties of the simplest possible electric field, namely, the field
about a hydrogen nucleus, we appear to find that this field is not a con-
tinuum but is strikingly discontinuous. As far as we are aware, the
electron cannot exist except in one of a series of levels, and whether
the idea of motion of an electron from one level to another has any
meaning is somewhat doubtful. As far as we can see, it disappears
from one level and appears at another. In this simple system what has
become of electric force? We might get something approximating
to the idea of force by dividing the energy difference between two
levels by the distance between the levels, but in sm:h a microcosmos

what assurance have we that the very idea of distance has any sig-
nificance? Should we not say perhaps that the distance between the
first and the third levels is 2, and that the difference between the first
and the seventh levels is 6?
Granting that the "field" about a positive particle has at least some
elements of discontinuity, and that perhaps this may be true also of
the field about an electron (provided that these two ideas are dis-
tinguishable), then since every electric field is a resultant of the fields
of these elementary particles, every electric field must have properties
of discontinuity. Instead of thinking then of an electric field as a
continuum, we should rather regard it as an intensely complicated
mesh composed of all the discontinuous elements due to the single ele-
mentary particles. Even if this view is correct, we need not for ordi-
nary purposes hesitate to use the equations of Maxwell any more than
we hesitate to employ the inexact methods of llydrodynamics in ordi-
nary problems.
An observer, moving rapidly past an electrostatic field, finds that
it is also a magnetic field, and if the electric field is discontinuous,
so is the magnetic. We need not abandon the brilliant idea of Maxwell
that light is an electromagnetic phenomenon, nor need we doubt the
approximate validity of his equations of the propagation of electro-
magnetic waves, provided that we consider them to have merely statis-
tical value. But when we consider the light emitted not from a great
aggregate of atoms but from a single atom, we may be sure that
this is something very different from that which is assumed in the un-
dulatory or electromagnetic theory. It probably bears to the electro-
magnetic wave a similar relation to that ~etween a molecule of water
and a quart of water. On the other hand numerous attempts to re-
turn to a corpuscular theory of light have hitherto failed to account
adequately for the phenomenon of interference. At present we may
say that we have no adequate theory of light.
The recognition that electric and magnetic fieilis are essentially dis-
continuous le~ds us to suspect that there is no such thing as a COn-
tinuous field of force; that a gradual acceleration accompanied by a
gradual increase in kinetic energy is something which does not exist
in nature. Rather we should consider that every system passes by
steps, which may be small but are nevertheless finite, from one energy
state to another.
Finally we might even suspect that space and time could better
be treated as discontinuous than as continuous, and represented by a
counting method rather than by the methods of a continuous geometry.
We might still call such a mathematical representation of space a
geometry, but it would be of a very different sort from any existing
geometry.-whether Euclidean or non-Euclidean, metrical or non-
metrical. Its elements would be nothing lIut points .and groups of
points, and a distance would always be an integral number. With
relation to some one paint, other points would be classified according
to whether tbey were separated from it by one step or two steps or
THE DISCONTINUITY OF PHI'SICO-CHEMICAL PROCESSES .65
n steps; and we might have a certain number of points in. the dass
removed by one step. another number in the class removed by two
steps, and so on.
I hoped at one time to be able to find such a network geometry
which, when the mesh was made exceedingly fine, would approximate
to the properties of Euclidean gCOJ:netry, but I am now convinced
that such an attempt is hopeles.. On the other hand, if we should
consider a single atom, we might be able to state that this atom by
itself determines a space which has just such properti.es of a network.
which we might represent by I central point, 4 points of the class
once removed, 9 points twice removed . 16 points in the third group,
and 25 in the fourth. If now we should admit a pair of e1ectrona
at each of these points we should have (ignoring the sub-groups at
Bohr) a representation of· the shells about an atom containing re-
spectively 2, 8, 18, 32 and 50 electrons. We should also find (once
more ignoring the sub-groups) that in such a geometry the distance
between two successive points would be quite without meaning, if
we could consider but one atomic system.
If, in some such way as I have crudely described, we could define
the space of a single atom, then general space might be regarded as
the composite of aU the spaces of all the atoms, and in this space we
could employ the ideas of extension, of distance, and the like, which
are used in Euclidean geometry; with the same sort of approximate
validity that we apply the principles of hydrodynamics to a system
containing a large number of molecules, or the principles of electro-
magnetics to a field generated by many elementary charges.
But it seems as we have proceeded that we have been getting
farther and farther away from physical reality into the domain of
metaphysical speculation, and it would be undesirable at present to
continue. Indeed in a periO<1 of transition such as the present we
must more than ever focus our attentlon upon our actual experimental
facts, and give less heed to those conventional abstractions of the
mind, such as force and fields of force, energy and the conservation
of energy, or even space and time. Some of these abstractions may
have to be abandoned as the conventional ether was abandoned after
the acceptance of relativity. Others may have to be modified, and
my chief purpose in writing the present section is not so much to
predict just how these modifications are to occur as it is to emphasize
the necessity of maintaining an openness of mind; so that, when the
solution of these problems, which now seem so baffling. is ultimately
offered, its acceptance will not be retarded by the conventions and
the inadequate mental abstractions of the past.
REFERENCES.
Abegg, 1904, Z. anorg. Chem., 39, 330. Eastman, 1922, I. Am. Chem. Soc., 44.
Abegg and Bodlander, 1899, Z. anorg. 438.
Chern., 453.
;?D, Einstein, 1907, Ann. det' Phys. [4]. ,22.
Adams, E. Q., 1916,1. Am. Chem. Soc., 180.
38, 1503. Edenmeyer, 1901, Amt., 316,43, 71. 75.
Arrhenius, 1887, Z. phys. Chem., 1,
631. Falk and Nelson, 1910, J. Am. Chem.
Aston, 192D, Phil, Mag. [6}, 39, 449, Soc., 32, r637.
6II. l'araday, 1833 and 1834, Phil. Trans.,
123.23; 124,77.
Baeyer, 1885, Ber., 18, 2277. Fliirschejm, 1902 and 1905, J. prakt.
Balmer, 1885, Wied. Ann., 25. So. Chem., 66, 321; 71, 497.
Bardwell, 1922, !. Am. Chern. Soc., 44, Franck and Hen:, 1913, Verh-. d.
2499· deutsch. Phys. Ges., I5, 34-
Berzelius, 1819, "Essai sur la theorie Fry, 19II, Z, physik. Chem., 76, 385.
des proportions chimiques et sur
l'infiuence chimique de l'electricite." Gibson and Argo, 1918, !. Am. Chern.
Paris. Soc., 40, 1327.
Bohr, I9IJ I, Phil. Mag. [6J, 26, I. Gibson and Noyes, 1922, I. Am. Chern.
Bohr, 1913 II, Phil. Mag. [6]. 26, 857. Soc., 44, 2091.
Bohr, 1921, Nature, 107, 104. Gomberg, 1900, 1. Am. Chem. Soc., 22,
Boltzmann, 1912, "VorJesungen iiber 757.
Gastheorie," Barth, Leipzig.
Born and Lande, 1918, Sitz. kg. preus. Helmholtz, 1881, I. Chern. Soc., 39, 277.
Akad., 19[8, 1048; Verh. d. deutsch. ,an't Hoff, 1875, "La chimie dans
phys. Ges., 20, 202, 210. l'espace." Rotterdam.
Brackett, 1922, Astrophys. !., 56, 154 Huggins, 1922, Science, 40, 679.
Bragg, 1920. Phil. Mag. [61. 40', 169. Hull, 1917, Phys, Rev., 10,661.
Bragg, 1922, Phil. Mag. [6J, 44, 433.
Bray and Branch, 1913, 1. Am. Chem. Jones, L. W., 1914, I. Am. Chern. Soc.,
Soc., 35, 1440. 36, Iz68.
van den Broek. 1914. Phys. Zeit., 14.
J2. Kekttle, 1858, Liebig's Ann., 106, 129-
Bury, 1921, J. Am. them.. Soc., 43, Kirchhoff and Bunsen, 1860 and 1861,
1602. Pogg. Ann., 110, 161; II3,337.
Korner, 1874, GaJI. chim., 4, 444.
Cannizarro, 1858, "Sunto di un corso Kossel, 1916, Ann. det' Phys., 49. 229·
di filosofia chimica." Geneva. Kratzer, 1920, Z. Physik., 3, 289.
c!e Chancourtois, 1863, "Vis tel1urique,
classement naturel des corps simples Langevin, 1904 and 1905, Campt. rend.,
au radicaux obtenu au moyen d'un 139, 1204; Ann. Chim. Phl'S., 5, 70.
systeme de classification helIcoidal et Langmuir, 1916, f. Am. Chtm. Soc., 38,
numerique." Paris. 222.
Crum Brown and" Gibson, 1892. I. Langmuir, 1919 I, J. Am.. ekem. Soc.,
Chem. Soc., (Jr, 367. 4I, 868.
Langmuir, 1919 II, I. Am. Chem. Soc.,
Dalton, 1808. "A New System of 4J. 1543·
Chemical Philosophy." London. Langmuir, 1920, J. Am. Chern. Soc .•
Davy. 1807. Phil. Trans., 97, t. 42, 274-
Dobereiner, 1816, Gilbert's Ann.~ 56, Lapwortb and Robinson, 1923, Farad.
JJ2. Soc., July meeting.
166
REFERENCES
Latimer and Rodebush, 1920, I. Am. Rutherford and Soddy, 1903. Pin'!.
Chem. Soc., 4z, 1419. Mag. l4], 5,445.
Le Bcl, 1875, Bull. Soc. Chim. [2], 23, Rydberg, 1890, Compt. reNd., 110. 394.
338. Rydberg, 1897. Z. arwrg. Chem .• 14. 66.
Lewis, G. N .• 1913, J. Am. Chem. Soc., (See also Marignac and Clarke,
35, 1448. "Constants of Nature," 5. 26a.)
Lt;_wis, G. N., 1916 I, !. Am. Chem. Rydberg, 1914, Phil. Mag. [6], 28, 144.
Soc., 38, 762.
Lewis, G. N., 1916 II, Proc. Nat. Acad. Schlenck and HoJz, 1917. Ber' F 50, 274.
Sci., 2, 586. 276.
Lewis, G. N., 1917, Science, 46, 297. Sidgwick, 1923, Nature, June 16.
Lewis, W. C. McC., 1916, 1. Chem. Sommerfeld, 1922, "Atombau und
Soc., 109, 796. Spektrallinien," Vieweg und Sohn,
Lowry, 1923, Trans. Farad. Soc. 18, Braunschweig.
Part 3, p. 3· Stark, 19I5, "Priuzipien der Atomdy-
Lyman, 1(}o4 and 19OO, Astrophys. !., namik, III, Die Elektrizitat im
19. 263; 23, 181. chemischen Atom," Hirzel, Leipzig.
Stern and Gerlach, r92l, Zeit. Physik.,
Marignac, 1860, Archives des Sc1enCL'S 8. no.
physiques et nature lies, Geneva, 9, Stieglitz, 1922, I. Am. Chtm. Soc .• 44,
97· [293.
Meisenheimer, 1913, Ann., 397, 273. Stock and others, 1912-1921, Ber., 45,
~.'lendeteeff, 186g, 1. Russ. Phys. Chem. 3539 (1912); Z. Elektrochem.. I9,
Soc., I, t. 779 (191J); Ber., 46, 1959, JJ5J
:Mcyer, L., 1870, Liebig's Ann. Suppl., (1913); Ber., 54 A, 142; 54 B, 531
M~'sei~~:
(1921).
1913 and 1914, Phil. Mag.
[6], 26, 1024; '7,70J. Thiele, 1899, Ann .• 306, 87.
Thomson, ]904. Ph.jJ. Mag., 7, :237.
Nef, 1904, I. Am. Chem. Soc., 26, 1549. Thomson, 1907, "The Corpuscular
Newlands, 1863, ekem. News, ;, 70. Theory of Matter," Scribner, New
Noyes and Lyon, 1901, J. Am. Ckem. York. .
Soc., 23, 460. Thomson, 1913, "Rays of Positive
Electricity and Their Application to
Parson, 1915. Smithsonian blSt. Publ., Chemical Analysis," Longmans,
p:;~r~'9i~: Ann. chim. phys. [81. 25, Green & Co., London.
Thomson, 1914, Phil. Mag. [6], 27.
P:s~en, 11)09, Ann. der Phys. [4], 27, T,;g;pe and Ingold, 1923, Intern.
F;;J~, 1908, Compl. rend., 147, 967;
Union Pure and Appl. Chem.
Townsend and Bailey, 1922, Phil. Mag.
148, 530.
PHrin, 1919, Ann. de phys., II, I. [6], 43, 593.
Fickering, 1897. Astrnphys. J., 5, 92.
Planck, 1901, Ann. der Phys. [4]. 4, VorHinder, 1922, Z. angew. Chem., 35,
249.
P;J:i:, 1815 and rSr6, Thoms(Jn's Ann.
Phil., 6, 321; 7, 111. Wahlin, 1932, Phys. Rev., I9, 173-
Weber, 1915, lahrb. d. Radioakt., 12,
Ramsay, 19Q8, I. Chem. Soc., 93, 774. 74·
Ramsay, 1916, Proc. Roy. Soc.# ;zA, Werner, 1905, I'Neuere Anschauungen
auf dem Gebiete der anorganischen
~kine, 1921, Proc. Roy. Soc. (A), Chertrie," Vieweg, Braunschweig.
Vlieland. 19II and 1914, Ann.~ 381, 200;
R:ie~: 1<)00, Phil, Mag. [5), 49, 539·
Rutherford. 19I1, Phil. Mag. [61. 21# \\I~::" :l¢:t!:~n.. tier Phys. [3], 58#
6li9> 662.
INDEX
Abegg, 30, 107 Bond, strained, '33, 135
Absorption bands, i61 Bond to heavy radicals. 144, 149
Acetylene, 126 Bonding pair, location, 83
Acctylide ion, 127 BOIlds, inadequacy, 156
Acid, definition, 138, 141, 142 Born, 56
Acids, dibasic, 139 Boron halides, 99
Adams, E. Q., 101, 139, 140 Boron hydrides, 95, 106, 123
Adsorption, 118 Boron, QuadtivaIent, too
Aldehyde·ammonia, 136 Boron tri-alkyl, 98, 124
Alternating properties, ISO Brackett series, 3B
Aluminum chloride, addition, 99 Bragg. 119, 121
Amide .'on, 137 Br.'tllch, 70; 84, 127, 128
Amine alkoxyl, decomposition, 135 Bray, 27. 70, 127
Amine oxides, 111, H2, 134 Broek, van den, 26
Amino-adds, 139 Brownian movement, 163
Ammonia synthesis, IS8 BUllsen, 35
Ammonium hydroxide, 110 Bury, 66
Ammonium ion, 87, 100, 106
Angular momentum, quantized, 48 Cannizzaro, 23
Argo, 148 Carhon bonds, 124
Argon, ionization potential, 147 Carbon monoxidt, I27
Arrhenius, 21, 157 Carbon, quinquivalent, 113
Associated liquids, no Carbonic acid, WI, 105
Aston, 19 Carboxyl, conjugation, 1$4
Atomic charges, 83 Carboxyl, substituted, 154
Atomic models, 29 Catalysts, 160
A torok numbers, 24, 25, 26, S5 Chancourtois, de, 23
Atomic theory, 17 Chemical affinity, source, 147
Atomic weights, integral, 18, 19 Chlora<:etic acid, 85, 139
Atoms, inner structure, 58 Chloramine as base, J 35
Atoms with two or more electrons, 52 Chlorine, positive, 131
Chlorouium ion, roB
Baeyer, 89, 93 Chromium kernels, 62
Bailey, 162 Cohalt kernels, 62
Balmer series, 36, 44 Collisions, elastic and inelastic, 46
Band spectra, 35 Color and loose electrons, 161
Bardwell, 34 Color and odd molecules, 161
Base, definition, 138, 141, 142 Color and variable kernel, 161
Benzene, 9:a, 146, 1S4 Combination principle, 37, 38, 46
Benzoic adds, substituted, 145 Conjugated double bonds, 90
BeryUium bond, 123 Conjugation and unsaturation, 92, Is6
Beryllium chloride, 98 Conjugation as a rearrangement, 91,
Beryllium, coordination, I23 '53
Berzelius, 20 Conjugation,. chemical and magnetic,
BodJander. 107 '411, '53
Bohr, 32, 37, 43, 44, 47, 56, .18. 76, 147 Continuum, IsS
Boltzmann, 39 Coordination number, 104. 1t4
Bond, 67, &, Coplanar atomic models, 52
Bond. between negative elements, 133 Copper kernels. 62
Bond, loose, 84. 149. 150, 159 Coulomb's law, 48. 50. 56
Bond, residual affinity, 149 Covalence, 105
169
170 INDEX
Crum Brown and Gihson rule, 144 Franklin, 20

Crystal structure, Il6 Frequencies, basic, 38
Crystals, hard and soft, 1'7 Friedel and Craft reactions, 99
Crystals, polar, 117 Fry, 71
Cubic atom, 29
Cyanide ion, IZl Geometry, network, 165
Gerlach, 58, 152
Dalton's theory. 17 Gibson, 148, 162
Davy, 20 Glutaeonic acid, 155
Diamagnetism, S3 Gold kernel, 61
Diamond, 116 Gomberg, 80
Diazo-compounds, J 56 Group of eight, 30, 33, 57, 66
Diazonium compounds, 156 Group of eight, magnetic stability. 149
Dichloretbane, 151
Dielectric constant, 53 Halogen bond, r56
Dipoles, S2 Heat content and temperature, 40
Discontinuity of fields, 164 Helium atom, 59
Discontinuity of matter, 17 Helium compounds. r22
Discontinuity of nature, 157 Helium, ionization, 147
llobereiner, 23 Helium series, enhanced, 49
Double bond, 88, 124 Helium spectrum, 37
Double bond, paramagnettc. 89. t48, Helmholtz, 2I, 35
Herz, 46
'S'
Dualistic theory, 20 Hildebrand, 99
Dualistic theory. modern, 67. 86 Hoff, van't, 20
Dulong and Petit law, 39 Holtz, lIZ
Huggins, 91, 109. 154
Eastman, 95, 124 Hull, 56
Einstein, 41, 42, IsS Hydrazine as base, 135
Electrochemical theory, 20, 137 Hydrides, 34, 121
EJectromers, lJI Hydrodynamics, 163
Electron arrangement in atoms, 55 Hydrofluoric acid, lIO, 129
Electron mass, 49 Hydrogen, active, 148
Electron mobility, 84. 143 Hydrogen as negative element, 33
Electron orbit as entity, 56 121, 136, 140
Electron orbit as magnet, 147 Hydrogen, bivalent, 109
Electron orbit, elliptic, 50 Hydrogen bond, log
Electron orbit, influence on other Hydrogen, diatomic, magnetism, 59
charges, 51 Hydrogen ion, 122
Electron orbit, radius, 4B Hydrogen kernel, 121
Electron orbits, separate, 56 Hydrogen, monatomic, magnetism, 58
Electron pair, 57, 66, 79. 80 Hydrogen peroxide, 130
Electron pair and magnetism, 58, 59, Hydrogen spectrum, 36, 44
'48 Hydrogen sulfide, 137
Electron. position, S6 Hydronium ion, 107
Electrons between two atoms, 75 Hydroxylamine as base, 135
Hydroxylamine tautomerism, loS
~~~;;nrev~l~~~ ~~o~hIS;ases, 162 Hypophosphorous acid, log
Equipartition law, 39
Erlenmeyer, 91 Ingold, r55
Ether and bromine, 107 Intramolecular ionization, 71
Ether and hydrochloric acid, 107 Iodine, positive, 132
Ethylene, 89, 124, 125, 126 Iodine, trivalent, 114
Iodottium ion, roB
Falk, 71 Ionization potential, 46
Faraday. 21 Ions,:' elementary. 61
Ferromagnetism, 53 Iron kernels, 62
Fluoborate ion, 86 Isocyanide ion, rz;
Fluorine and oxygen, lJO Isotopes, 19
Fliirscheim, 145. 156
Franck, 46 Jones! lIt
INDEX.
K-Ievel,45 Nickel kernels, 62

Kekule, 20 Nitric add, WI, 105
Kernel, 29, 30, 57, M Nitric oxide, 128
Kernel, variable, 61 Nitrogen bonds, 124
Kirchhoff, 35 Nitrogen dioxide, 129
Korner, 92 Nitrogen, molecular, 126
Kassel, 33, 7'1, I2$' Nitrogen pentoxide. 129
Kratzer, 121 Nitrogen, quadrivalent, 87, III
Nitrogen, quinquivaJent, 113
L-level, 45 Nitrogen trichloride, 73, 85, 132
Lande, 56 Nitrogen trichloride as base, 135
Langevin, 53 Nitrogen trioxide, 129
Langmuir, 79, 8;, 97, 104, II;, 12; Nitroso-eompounds, uS
Lapworth, 146 Nitrous acid. 141
Latimer, 101, 109, 125, 140 Nitrous oxide, 129
Le Bel, 20 NO" 128
Lewis, G. N., "9, JJ, 57, 61, 79, &4 Noyes, W. A., 73. uS
Lewis, W. C. MtC., 157 Noyes, W. A., Jr., 162
Light, emission and absorption. 35. 44, N ueleus, 20, 26, 4S
48, 57, 160 Nucleus and magnetism, 58
Line spectra, 35
Lithium, 122 Octet, 79, see Group of eight
Lithium, ionization, 147 Octet theory. 97
Long periods, 60 et seq. Odd molecules, So, 148
Lowry, 130 Olson, 158
Lyman series, ,38, 44, 47 Onium compounds, 106, 130, 134
Orienting groups, 144, 146
M-level, 45 Ortho-acids, 105
l\1aeser, 112 Osmium tetroxide, 86
M agnetic fi~ld, 52 Oxidation and reduction, 69
Magnetic fields, neutralization, 149 Oxonium, 107, 139
Magnetic moment, 53 Oxygen, paramagnetism, 89
,j1agnetic moment, atomic, 58 Oxygen, univalent, 86, 135
Magnetic moments, integral, 148. 152 Ozone, 130
Magnetic permeability, 53
J\,'Iagnetic phenomena, 52 Paramagnetism, 53, 152
~lagnetic sQsceptibilities, atomic, 152 Parhelium, 122
Magnetism and molecular distortion, Parson, 32, 77
ISO Partial valence, 91
Magnetism and unsaturation, 151 Pascal, 92. lSI, 153. 156
Magnetochemical theory, 147 et seq. Paschen series, J8
Magneton. 32 Perch lone acid, 86, rcB
Maleic and fumaric acids, 144 Per-jodie acid, 106, 113
Manganese kernels, 62 Periodic Jaw, 23
Marignac, 19 Periodic table, 28
Maxwell, IS8 Perrin, 18, 157
Meisenheimer, III Phenyl sulfuric acid, 70, 85
Mendeleeff, 23, 30 Phlogistication, 6g
Mercury bonds, 115 Photochemical reactions, 157
Metaphosphoric acid, 101 Photo~electric law, 42
Meyer, 23 Phosphate ion, 86
Molecular dimensions, 119 Phosphonium ion, 106
Molecules in a solid, II; Phosphorous acid, 109
Moseley, 26, 45 Phosphorus acids, 140
Multiple bonds, limitation, 94 Phosphorus, no doubJe bond, 95
Phosphorus pentachloridet 101, 113
Nef, I:z6, 128 Phosphorus, quacirivalen4 109
Negative groups, 85 Pickering series, 37. 49
Negative states, J31 Planetary atom, 32, 5S
Nelson, 71 Planck, 4 1, 4S
Newlands, '3 Planck constant, 41, 44, 57, 161
iNDEX
Platinous salts, complex. 115 Strain theory, &;, 93

Polar and nonpo1ar substances, 21, 71 Strength of electrolytes, 129, 135, IJ8
Polar effect through chain, 139 Structural chemis.try, 21
Polar number, 70, 104 Structural formulae, 6;
Polarity, 83, B4. 128, IJ3 Subliming substances, 99
Polarity and graphical formula, 133 Sulfate ion, 86
Polarity and tautomerism, 134 Sulfonium ion, 106
Polarity, alternatin~, 145, 146 Sulfur hexa.fluoride, 102, Il3
Positive and ncgahve radicals, 142 Sulfur, no double hand, 95
Positive states, 131 Sulfur trioxide, 100
Potassium, ionization, 147 Sulfuric a<:.id isomers, 100
Prout, I8 Sulfurous acid, 140
Quadratic groups, 27 Tautomerism, mohile hydrogen, 100,

Quadruple hond, 127 108, 128, 130, 136
Tetrachlor~ethylene, 156
Quantum of energy, 41 Tetrahedral structure, 82
Quantum theory, 39, 132, 157 Thallium kernel, 61
Quantum theory, future, 162 Thiele. 91
Thomson, 19, 21, 3D. ;1, 74
Radiant energy, distribution, 40 Thorium, 27
Rankine, I 19 Thorpe, 155
Ramsay, 74. 147 Titanium kernels, 61, 62
Rare earths, 6J, 65 Towns('nd, 162
Rare: gases, 30, 33 Transition elements, 62, 63
Rayleigh, 40 Transition elements, magnetism, 64
Residual affinity. 92, 149 Triphenytmethyl, 144, 148
Resonance potential, 46 Trjphenylmethyl carbinol, 144
Richardson, 98 Triple bond, 93. 126
Robinson, 146 Triple bond, magnetism, 153
Rodebush, 101, 109, 125, 140
Rule of eight. exceptions. 97 Unity of matter, 18
Rutherford, IS, 26, 32, 55 Unsaturation, 88, 124, 126, 149
Rydberg, 19, 27. 37, 39, 122 Unsaturation and loose bonds, 90, 143
Rydberg constant, 37, 49 Umaturation, distribution, ISO
Rydberg's ordinal numbers, 24 Uranium, 27
Uranium hexafluoride. 103
Schlcnck. 112 Uranium tetrachloride, 103
Secondary bonds, 115 Urey, 119
Secondary valence shells, 1I 5
Shells, concentric, 30, 57 Valence, defined, 104, lIB
Shells, electron capacity, 58 Valence in condensed systems, Il6
Sidgwick, 123 Valence shells, secondary, 102
Si1icate 10n, 86 Vanadium kernels, 62
Sillcon, np double bond, 94- VorUinder, 83
Soddy, .8
Sodium in ammonia, 148 Wahlin, 100
Sodium methyl, 9B Walden Inversion, 113
Sommerfeld, 147 Water oomplexes, no
Space and time. 164 Water. ionization, 129
Spectral series. 35 Water molecule', unsy-mmetric, 8:3
Stark, 75 Weber, 64
Static atom, 55 ~ Werner, 68, 86, 87, II4, ns, 137, 140
Stern, 58, 152 Wieland,80
Stewart, 112 Wien,40 ,
Stieglitz, 71, I4S. lSI
Stock, g8
Storch. 158

2번

Uploaded by

Copyright:

Available Formats

2번

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2번

Uploaded by

Copyright:

Available Formats

VALENCE

American Chemical Society

Th e CH EM I CAL CATALOG COMPANY, I nc.

All Rights Reserved

American Chemical Society Series of

AMERICAN CHEMICAL SOCIETY

Scientific Series:- Technologic Series:-

The CHEMICAL CATALOG COMPANY, Inc.

CHAPTER 1. THE ATOMIC THEORY 17

THE DISCONTINUITY OF lVfATTER. 17. The ,Vark of Dalton,

CHAPTER II. TIlE PERlOllIC L\w .\.:\1) THE CIIE:\fIST'S PiCTURE

CHAPTER III. SPECTRAL SERIES A::\'1J THE PIIYSTCIST'S VlE\,·; OF

ClIAPTBR IV. REC01\~CILLI\TION OF THE Two VIEWS; THE AR-

CIL<\J'TER V. TllE t~]'OO:.";' OF A1'OMS; TIlE MODERN DLTALISTIC

CILA,I'TER Y1. THE l\~E\y TllEORY OF V.\LENCE; THE CHEMICAL

CHAPTER VII. DOUBLE AND TRIPLE BONDS • 88

CUAPTER VIII. EXCEPTlOl"S TO TIlE RULE OF EIGHT 97

CHAPTER IX. 'VALENCE AND COORDI'K/\TlO!I.~ Nl1MHER 104

A Redefinition of Valence. J04. The Normal Valence of

CHAPTER X. COMPOUNDS OF ELEMENTS \YITH SMALL KERNELS II9

Molecular Radii, II9. HYDROGEN AND IIELICM, J2J.~

CHAPTER XI. ELEMENTS IN POSITIVE AND NEGATIVE STATES 131

Criticism of our Present Terminology, 131. THE PROBLEM

CHAPTER XlI. REMNANTS OF THE ELECTROCiIEl\ftCAL TUEORY 137

CUAPTER XIII. THE SOURCE OF CHEMICAL AFFINITY; A 1\1AG-

CONJl'(;ATlON. ]53. Chemical Conjt1~ati()11 h l\ta~n('tit

CUAPTER XIV. THE DISCONTINTITY OF PHYSICo-CHEMICAL

VALENCE AND THE STRUCTi"T'D''iii;;;;;;iiiiii"-

The Unity of Matter.

numbers, an instinctive tendency toward the rounding off of uncertain

even if we include those capable of exact mathematical statement, has

the other hand a relatively non-polar substance like diatomic hydrogen

If we glance at the table of the elements we see that there is a differenoe

FIG. I.-X-Ray Emission Lines.

halogens. So also the positions of argon and potassium and of cobalt

FIG. 2.-Atomic KU011){'r and X-Ray Frcqu('ocy.

Thus the relation of magnesium to zinc is not clearly indicated~ and

Some Atomic Models.

FIG. 3.-Lewis: Memorandum of 1902.

although it contained certain crudities, I have ever since regarded as

(2) A neutral atom of each element contains one more electron

where another ring is formed, and so on. As an illustration of the

is entirely similar to the one offered by Abegg, although unlike the

:I A brief description of some of the elementary principles of magnetism will

Kossel, "'Cber Molekiilbildung aIs Frage des Atomhaus," and onc by

pressed ill Figure I) assumed these groups to constitute concentric

thermal action Of by electric discharge, are then the source of light;

where A is the wave length and n is anyone of the whole series of

Modern refinements of the methods of spectroscopy have greatly in-

OOmber of basic frequencies. In order to il1ustrate this point we may

Now it is evident that a 1ine of anyone of these series may be

therefore the frequency of every line can be regarded as the difference

\Vhen we 'turn to other elements than hydrogen we find a condi~

differences between certain basic frequencies, in complete accordance

The Quantum Theory.

FIG. 7.-Hcat Content of Copper,

except that which is due to the translational motion of its molecule

fully by Planck (H)OI). This curve approaches a straight line, as

FIG. 8,-Density of Radiant Energy (Between Two Given Frequencies).

So the second line of the Balmer series, which is V2-V,,. is supposed