THE
PHYSICAL SOCIETY OF LONDON,
REPORT
ON
-SERIES IN LINE SPECTRA.
BY
A. FOWLER, A.R.C.S., F.INST.P., F.R.S.,
Professor of Astrophysics, Imperial College of Science and Technology
South Kensington, London.
Price, 12/6. Post Free, 13/-.
Bound in Cloth, 15/6. Post Free, 16/-.
LONDON :
FLEETWAY PRESS, LTD.,
3-9, DANE STREET, HIGH HOLBORN, W.C. i.
1922.
LONDON :
THE FLEETWAY PRESS, LTD.,
DANE STREET, HOI^ORN, W.C. i.
PREFACE
ALTHOUGH the spectra of elements and compounds were studied in the first
instance chiefly as providing a powerful means of chemical analysis, it has
long been recognised that a spectrum must contain an important clue to
the structure and modes of vibration of the atoms or molecules which produce
it. Spectra, however, are most frequently very complex, and there could be but
little hope of progress in the direction indicated if it were not possible to discover
laws governing the distribution of the lines or bands of which they are composed.
The search for such laws has attracted many workers, and organised systems of
lines which can be approximately represented by simple formulae have been identified
in the spectra of many elements and compounds. The recognition of these regulari-
ties has naturally played a fundamental part in the development of theories of the
origin of spectra and of the constitution of atoms and molecules. The analysis of
spectra has thus become one of the main objects of modern spectroscopy, stimulating
the experimentalist to the extension of existing data, and providing material in a
form suitable for the theoretical investigator.
My purpose in the present report has been to give a comprehensive account
of the development and present state of our knowledge of the regularities in spectra,
as deduced from the spectra themselves, with but little regard to theories of their
origin. The report is in two parts, the first of which gives a general account of
spectral series, excluding those which occur in band spectra, while the second is
intended to include the most authentic experimental data available in April, 1921.
It is hoped that the tables of series lines, together with the references to lines which
have not yet been classified, will suggest and facilitate further investigations. The
system of numeration of the series lines which has been adopted is that of Rydberg and
Hicks, but if it should be found convenient to modify this numeration on theoretical
grounds there should be no difficulty in making the desired alterations.
The spectra dealt with are those obtained by optical methods, extending from the
infra-red to the extreme ultra-violet. The emissions of higher frequencies which
have been revealed in the study of X-ray spectra do not form extended series of the
kind met with in optical spectra, and have accordingly not been considered.
I am indebted to Professor F. A. Saunders, of Harvard University, for much
useful help during the preparation of the report, and especially for his kindness in
supplying important observational material in advance of publication. My thanks
are also due to Col. E. H. Grove-Hills, F.R.S., for the photograph reproduced in
Fig. 6 ; to Dr. A. S. King, of the Mount Wilson Observatory, for Plate III. b ; and
to Mr. N. R. Fowler for the negatives from which Plate IV. was prepared. I also
have pleasure in acknowledging the valuable assistance which has been rendered
by Dr. D. Owen, Secretary of the Physical Society, and Mr. H. Dingle, B.Sc., D.I.C.,
in reading the proof sheets.
A. FOWLER.
IMPERIAL COLLEGE,
SOUTH KENSINGTON, LONDON.
February, 1922.
^8! 1.1)5
CONTENTS.
PART I. GENERAL ACCOUNT OF SERIES.
CHAPTER PAGE
I. OBSERVATIONAL DATA 1
Wave-lengths and wave-ii umbers. Correction to Vacuum. Arc and Spark
Spectra : Enhanced lines. Intensities and Characters of Lines. The Schumann
and Infra-red regions. Sources of data.
II. HISTORICAL NOTE 7
1869-1879. The spectrum of hydrogen. Schuster's conclusions. I/aw of constant
separations. B aimer's law for hydrogen. First results of Kayser and Runge. The
work of Rydberg. Kayser and Runge's formula.
III. CHARACTERISTICS OF SERIES 13
General formula for hydrogen. The four chief series of other elements. Relations
between the chief series. The Rydberg-Schuster law. Runge's law. Abbreviated
notation. Doublet systems. Triplet systems. Satellites. Negative wave-numbers.
The combination principle. Enhanced line series. Identification of series. Spectra
with constant differences.
IV. RYDBERG'S FORMULA 27
The Rydberg constant. The constants jz and A . Rydberg's interpolation table.
The order-numbers of the lines. Enhanced line series. Rydberg's special formula.
V. OTHER SERIES FORMULA 31
Runimel. Ritz. Lohuizen. Mogendorff -Hicks. Paulson. Johanson. Ishiwara.
VI. "ABNORMAL" SERIES 39
Intensities. Satellites. Spacing of lines.
VII. SPECTRA AND ATOMIC CONSTANTS 43
General relationships. Limits and atomic weights. Limits and atomic volumes.
Doublet and triplet separations. Homologous lines and atomic weights. Separa-
tions and atomic numbers. Conclusions.
VIII. THE WORK OF HICKS 51
Atomic volumes. Atomic weight term. The " oun." "Collaterals." Constitu-
tion of diffuse series. Links. Summation series. Independent determination of
atomic weights.
IX. APPLICATIONS OF BOHR'S THEORY 59
The spectrum of hydrogen. Ionised helium. Arc spectra. Spark spectra.
Graphical representation of- series systems. lonisation potentials in relation to
series. Spectra and the periodic table of the elements.
APPENDIX I 76
Calculation of formula constants. Successive approximation. Hicks's method.
The differential method. Determination of limits in special cases.
APPENDIX II. 80
TABLES FOR COMPUTATIONS.
I. Corrections to reduce wave-lengths on Rowland's scale to the international scale.
'II. Correction to vacuum of wave-lengths in air at loC. and 760 mm.
I!A. Correction to vacuum of wave-lengths in infra-red.
III. Rydberg's interpolation table (revised) .
PART II. TABLES OF SERIES LINES.
CHAPTER PAGE
X. EXPLANATION OF TABLES 87
XI. HYDROGEN AND HELIUM 89
XII. GROUP I A., THE ALKALI METALS 96
XIII. GROUP IB., COPPER, SILVER AND GOLD 109
XIV. GROUP HA., THE ALKALINE EARTH METALS 115
XV. GROUP HB., ZINC, CADMIUM AND MERCURY 139
XVI. GROUP IIlA., SCANDIUM, YTTRIUM AND THE RARE EARTHS 152
XVII. GROUP Ills., THE ALUMINIUM SUB-GROUP ... 155
XVIII. ELEMENTS OF GROUPS IV. AND V. 163
XIX. GROUP VI., OXYGEN, SULPHUR AND SELENIUM 166
XX. ELEMENTS OF GROUPS VII. AND VIII 173
XXI. THE INERT GASES 174
INDEX OF AUTHORS 181
LIST OF PLATES.
PLATE
I. ARC SPECTRA OF SODIUM AND LITHIUM.
II. ARC SPECTRA OF CADMIUM, ZINC AND MAGNESIUM.
III. (a) TRIPLETS OF ZINC, CADMIUM AND MERCURY.
(b) SATELLITES IN FUNDAMENTAL TRIPLET OF BARIUM.
(c) SHARP, DIFFUSE, AND FUNDAMENTAL TRIPLETS OF CALCIUM.
IV. ARC AND SPARK SPECTRA OF MAGNESIUM, CALCIUM, STRONTIUM
AND BARIUM.
V. SPECTRA OF HELIUM AND IONISED HELIUM.
FIGURES IN TEXT.
FIG. PAGE
1. THE SPECTRUM OF HYDROGEN 8
2. THE THREE SERIES OF HYDROGEN 14
3. THE FOUR CHIEF SERIES OF LITHIUM 17
4. DIAGRAM ILLUSTRATING ARRANGEMENT OF SATELLITES 20
5. (a) SERIES IN WHICH P(l) is POSITIVE ; (b) SERIES IN WHICH P(l) is
NEGATIVE 22
6. PHOTOGRAPH SHOWING AN INVERTED DIFFUSE DOUBLET OF IONISED
CALCIUM 22
7. CURVE OF /* +f(m) FOR THE PRINCIPAL SINGLET SERIES OF CALCIUM. . , . . '. 42
8. THE SHARP SERIES OF Mg, Ca, Sr, Ba, SHOWING INFLUENCE OF ATOMIC
WEIGHT 44
9. THE BOHR ORBITS FOR HYDROGEN 61
10. COMPARISON OF EXTENDED PICKERING SERIES WITH THE BALMER SERIES
OF HYDROGEN ' , ... 63
11. TERMS OF THE HYDROGEN SPECTRUM 66
12. TERMS OF THE SPECTRA OF Na, Mg, Mg + , Sr, Sr + 6T
PART I.
GENERAL ACCOUNT OF SERIES.
CHAPTER I.
OBSERVATIONAL DATA.
WAVE-LENGTHS AND WAVE-NUMBERS.
The study of spectral series calls for a precise acquaintance with the nature of
the available data, both as to the positions and characters of the lines involved.
The position of a line in the spectrum is most generally indicated by the wave-
length (A) of the vibrations which produce it. The unit of wave-length is the
Angstrom Unit, or " angstrom," as it is now beginning to be called. It was intended
to equal 10* 10 metre, and is accordingly often called the "-tenth-metre." It is also
equivalent to 10" 8 cm., or 0'0001/j, where ,a is the micron, or thousandth of a milli-
metre. Wave-lengths in the visible spectrum range from about 3,900A to 7,600A
(A being the modern abbreviation for the angstrom). For the long waves in the
infra-red, however, /i is often taken as the unit, so that Al 2,500 A, for example,
would be indicated by 1'25/j.
The wave-length scale was introduced by Angstrom in 1869, and until about the
year 19QO, wa\e-lengths were referred to his determinations as standards ; they were
meant to represent the wave-lengths in air at 16C. and 760mm. pressure. Later
work, however, proved that Angstrom's values were about 1A too low, and, about
1896, the scale was superseded by that of Rowland, which referred to wave-lengths
in air at 20C. and 760mm. Still more recently it has been found that Rowland's
scale was based upon erroneous values for the D lines of sodium, besides being affected
by other small errors, and Rowland's scale is now being gradually replaced by the
" International " scale. The latter is founded upon interferometer determinations
of the wave-length of the red line of cadmium, which is much superior to the sodium
lines in point of sharpness. The wave-length adopted for this line by the Inter-
national Solar Union* is 6438 '4696, being the value determined by Benoit, Fabry,
and Perot, and in close accordance with a previous measurement by Michelson.
The precision of this value for the standard line is such that the unit of wave-length
which it defines can differ but little from 10~ 10 m., but to avoid all misunderstanding the
unit of wave-length defined by the cadmium line has been called the International
Angstrom ; and is indicated in tables by the letters " LA." ; we thus have
_Wave-length of red Cd line in dry air at 15C., 760 mm., with g=980'67
6438-4696
The majority of published wave-lengths, however, have been expressed on
Rowland's scale, and if new determinations have not become available, it is necessary
to reduce the Rowland values to the international scale in any precise work on series.
From a comparison of the two scales, Kayserf has derived the corrections which are
shown in Table I. These corrections include the small differences depending upon
the differences in the standard temperatures of the two scales (20C. for Rowland, and
15C. for the international scale).
In connection with spectral series it becomes important to specify the positions
* Trans. Int. Sol. Union, 2, 20, 28 (1907). (The Solar Union is now absorbed into the Inter-
national Astronomical Union.)
fHandbuch der Spectroscopie, 6, 891 (1912).
2 Series in Line Spectra. CHAP. i.
of lines either in " oscillation-frequencies," or by " wave-numbers." The most
fundamental figures are the oscillation-frequencies, since these are not changed when
the medium is changed. But the determination of frequency requires an exact
knowledge of the velocity of light, and it is more convenient to use the wave-number,
or number of waves per centimetre ; thus
10 8
Wave-number = v = -
/ in Angstroms
CORRECTION TO VACUUM.
Wave-numbers as well as wave-lengths vary with the medium in which the
vibrations are propagated, and it is therefore necessary to reduce them to standard
conditions. They are accordingly reduced to their values in vacuo, and when thus
corrected the wave-numbers are strictly proportional to the oscillation-frequencies.
This correction can be made when the refraction and dispersion of air have been
determined with sufficient accuracy. Thus
bac. = /* ^air > or A A = A vac . A a/ > = A a; > (^ 1 )
where /^=the refractive index of, air.
The refractive index, of course, varies with the density, and therefore with the
temperature and pressure, of the air. Kayser and Runge have given a formula
indicating the value of p for air at 760mm. pressure, and temperature 0C., and have
derived a table of corrections to vacuum applicable to wave-lengths determined in
air at 20C. and 760mm. pressure.* The most recent observations have been made
at the Washington Bureau of Standards by Meggers and Peters.f who give the
following formulae for the refractive indices of dry air at 760mm. pressure, and tem-
peratures 0C., 15C. and 30C., \ being expressed in angstroms :
0C.: u - 1
30'C.: ( ,
These observations have been utilised in the construction of a table showing
the corrections which must be applied to wave-lengths and wave-numbers measured in
air to convert them to their values in vacuum. An extract from the Washington
tables is given in Table II., indicating the amounts to be added to wave-lengths in
dry air at 15C. and 760mm., and therefore directly applicable to wave-lengths on
the international sca]e.J The conversion to wave-numbers can readily be effected
by the use of a table of reciprocals with seven-place arguments.
When the wave lengths are expressed on Rowland's scale, which refers to 20C.,
the simplest procedure is to reduce them to the international system by means of
* Handbuch, 2, 614.
t Scientific Papers of the Bureau of Standards, Washington, No. 327 (1918).
% Some of the figures have been slightly amended in Astrophys. Jour., 50, 56 (1919).
Observational Data. 3
Table I., and then to correct them to vacuum bv adding the corrections given in
Table II.
The corrections of infra-red wave-lengths to the international system and
to vacuum are subject to some uncertainty. For wave-lengths greater than
10,OOOA, however, the correction to the international scale has but little in-
fluence on the wave-numbers. Thus, at A10,000 the correction increases the wave-
number by 0-4, at A20,000 by 0'2, and at A40,000 by 01. Corrections to vacuum
may be determined approximately by extrapolation of the Washington formula
for the refractive index of air. (See Table Ha.)
ARC AND SPARK SPECTRA : ENHANCED LINES.
The investigation of spectral series often requires a knowledge of the behaviour
of the lines when produced under different conditions of excitation. The
spectra of metallic elements are most frequently obtained by the use of the electric
arc or by the condensed discharge from an induction coil, and the spectrum is usually
different in the two cases. (See Plate IV.) On passing from the arc to the spark
it often happens that some of the lines are diminished in relative intensity, while
others become brighter, and new lines frequently -make their appearance in the
spark. Lines which are relatively enhanced in brightness on passing from the
arc to the spark, or which only occur in the spark spectrum, were called Enhanced
Lines by Lockyer, and this name has been generally adopted. Lines which appear
in the arc and tend to diminish in intensity in the spark are then distinguished as
Arc Lines. The term " arc lines," it will be seen, does not necessarily include all
the lines which appear in the arc spectrum ; and, similarly, the spark spectrum
most frequently includes some surviving arc lines as well as enhanced lines.
It is convenient to distinguish at least three classes of enhanced lines : (I.) En-
hanced lines like the H and K lines of calcium, which are quite strong in the arc
(and appear with greater intensity in the spark) ; (II.) Lines which only appear
with feeble or moderate intensity in the ordinary arc, such as the enhanced lines oi
iron and titanium ; (III.) Lines which do not appear in the ordinary arc, but are
strongly developed under spark conditions, as in the case of the well-known mag-
nesium line A4:,481. It would thus seem that while the energy of the arc is in some
cases sufficient to develop the enhanced lines, the greater energy of the disruptive
spark is required to give rise to them in the case of some of the elements. In other
words, different elements respond differently to a given stimulus.
It should be mentioned that enhanced lines are also often found to appear in a
region close to the poles of a metallic arc, and when the arc is passed in an atmosphere
of hydrogen, or in a vacuum,* they frequently become very pronounced.
Similar variations of the line spectrum are also observed in the case of many
gases, including helium, oxygen, and nitrogen, when the electric discharges which
excite them are varied in intensity. From analogy with metallic spectra,
certain lines which thus appear in the spectra of gases under the action of powerful
discharges may quite properly be classed as enhanced lines. In some cases, a
succession of spectra appear as the intensity of the discharge is gradually increased.
Oxygen, for example, gives a spectrum corresponding to the arc spectrum when the
discharge is feeble, new lines appear with a moderate increase in the intensity, and
still others when the strongest discharges are passed through the gas ; the three
* Fowler and Payn, Proc. Roy. Soc., 72, 258 (1903).
B 2
4 Series in Line Spectra. CHAP. r.
different classes of lines are then conveniently distinguished as arc, spark and '' super-
spark," or as 0i., On., and 0m.* Silicon shows four such stages.
As might be expected, arc lines and enhanced lines have not been found to be
associated in the same family of series.
INTENSITIES AND CHARACTERS OF LINES.
In the investigation of series spectra, it is important also to take into account
the intensities and physical characteristics of the lines involved. A convenient
standard scale of intensities has not yet been introduced into spectroscopic tables,
and tabulated intensities are, for the most part, merely rough estimates on an arbi-
trary scale, in which 10 represents the strongest and 1 the weakest lines. This
range, however, is often too restricted, and in some tables, notably those of Exner
and Haschek, the very strongest lines are represented by the higher numbers 15,
20, 30, 50, 100, 200, 500 and 1,000. In order to extend the scale in the opposite
direction, some observers also follow Rowland's convenient plan of indicating very
faint lines by 0, 00, 000, and 0000, the latter being at the limit of visibility. A
method of estimating intensities on an absolute scale has been devised by Nicholson
and Merton.f but its use has so far been restricted to the spectrum of helium.
While some lines are sharp and well-defined, others may be shaded on one or
both edges, and others again, especially in arc spectra, may be reversed. These
different appearances are usually indicated in spectroscopic tables by the addition
of letters to the numbers showing the intensities. There is, unfortunately, a con-
siderable diversity in the notations adopted by different observers, and it is usually
necessary to rely upon an author to describe the symbols employed in any particular
case. Some of the principal systems which have been adopted are as follows :
Exner and This
Character. Watts. Kayser. Haschek. Report.
Sharp s ... s* ... ... s
Nebulous or diffuse n . .'. j ... + ... n
Broad b ... wf ... br% ... b
f Broad, but sharp on red edge ... b r ... ... ...
\ Diffuse towards violet ... ... ... v, or uv . . . v ... v
/Broad, but sharp on violet edge ... b v ... ... ...
\ Diffuse towards red ... ... ... ... r, or ur . . . r ... r
Reversed r ... R ... ... R
Double ... d ... d ... d
* Scharf. f Unscharf. { Breit. Umgekehrt.
The symbols in the last column will be adopted in this report, and it is to be
understood that the absence of any symbol indicates that the line is of ordinary
sharpness, but not specially sharp.
In a reversed line there is a broad bright line, diffuse at the edges, which is
produced by the denser vapour at the core of the arc or spark used as the source of
light, and a narrow absorption line superposed upon this which originates in the
cooler and less dense vapour in the outer envelope. The reversal is sometimes
unsymmetrical.
* Fowler and Brooksbank, Monthly Notices, R.A.S., 77, 511 (1917).
t Phil. Trans., A, 217, 242.
Observational Data. 5
It should be noted that lines which are ordinarily diffuse in an arc spectrum
can usually be obtained as sharp lines by passing the arc in a vacuum. There are,
in fact, examples of series which might never have been recognised as such if reliance
had been placed upon observations of the arc or spark in air, as the more refrangible
members are sometimes diffused to the degree of invisibility.
Lines which belong to the same series are usually similar in character, but an
apparent exception has been noted by Royds in barium ;* in this case the lines of a
triplet in the yellow appear to be widened unsymmetrically towards the red, while the
more refrangible triplets are shaded towards the violet. The corresponding triplets
of calcium present a le:s extreme exception, the blue triplet being sharply denned,
while succeed.ng members are diffuse towards the violet ; the associated infra-red
triplet may possibly be shaded towards the red.
THE SCHUMANN AND INFRA-RED REGIONS.
The ordinary instruments which are employed in spectroscopicwork do not permit
the investigation of the whole range of the spectrum. Spectroscopes with prisms
and lenses of glass only serve for a small part of the ultra-violet, and a small part of
the infra-red, in addition to the visible spectrum. When quartz is substituted for
glass the range may be extended to about A 1,850 in the ultra-violet, but for shorter
wave-lengths special arrangements become necessary. This part of the spectrum
was first investigated by Schumann, and is commonly called the Schumann region.
In the first instance, fluorite was substituted for quartz, and on account of the opacity
of air for the short waves the whole apparatus, including the photographic plate,
was placed in an exhausted air-tight case. The ordinary gelatine plates being strongly
absorbent for short waves, special " Schumann plates " with a very thin coating of
gelatine are also necessary. Concave gratings have been successfully employed by
Lyman and others, and have permitted observations beyond the region for which
fluorite is transparent. Details of the instruments and methods of work have been
given by Lyman in his book on " The Spectroscopy of the Extreme Ultra- Violet. "t
Observations have now been extended as far as A 584A by McLennanJ and to
22oA by Millikan. Wave-lengths in this region are usually tabulated as observed,
and no correction to vacuum is required in the calculation of wave-numbers. It
should be noted that since dv= - 10 8 /A 2 .^A, errors in the wave-length are greatly
multiplied in the conversion to wave-numbers in this region as compared with
the less refrangible parts of the spectrum.
At the red end of the spectrum, direct photographs ori plates stained with
dicyanin have been obtained by Meggers as far as HO,OOOA.|| The further infra-
red is investigated by thermo-electric methods, emplojnng spectroscopes having
optical parts of rock-salt or making use of gratings. Extensive work with special
reference to series lines has been carried on in this region by Paschen^I and by Randall.**
The extension of observations into the extreme ultra-violet and infra-red has
been of great value in the elucidation of the structure of spectral series, as will
appear in due course.
* Astrophys. Jour., 41, 154 (1914).
t Longmans (1914).
: McLennan, Proc. Roy. Soc., A., 95, 238 (1919).
Astrophys. Jour., .52, 47 (1920).
|| Scientific Papers of the Bureau of Standards, Washington. Numbers 312, 324, 345, &c.
^ Ann. d. Phys., (4) 27, 29, 33 (1908-10), and other Papers.
Astrophys. Jour., 34, 1 (1911) ; 42, 195 (1915) ; 49, 42, 54 (1919).
**
6 Series in Line Spectra. CHAP. i.
SOURCES OF DATA.
The following references to the principal collected tables of wave-lengths and
photographs of spectra may be usefully appended to this chapter.
(1) W. MARSHALL WATTS : " Index of Spectra," with numerous appendices.
(Heywood, Manchester.)
(2) F. EXNER and E. HASCHEK : " Wellenlangen Tabellen fur Spectralana-
lytische Untersuchungen auf Grund der Ultra- Viol etten Funkenspektren
der Elemente." (Leipzig and Wien, 1902.)
(3) F. EXNER and E. HASCHEK : " Wellenlangen Tabellen . . . Bogen-
spektren." (1904.)
(4) F. EXNER and E. HASCHEK : " Die Spektren der Elemente bei Normalen
Druck." I., Hauptlinien der Elemente und Codex der Starken Linien
im Bogen und Funken. II., Die Bogenspektren. (Leipzig and Wien,
1911.)
(5) A.HAGENBACH and H. KONEN : ."Atlas der Emission Spectra." (Jena,
1905.) English edition by A. S. King. (W. Wesley & Son, London.)
(6) J. M. EDER and E. VALENTA : " Atlas Typischer Spektren." (Wien,
1911.) |
(7) H. KAYSER : " Handbuch der Spectroscopie," Vols. V. and VI. (Leipzig,
1910, 1912.) (These include practically all the measures to the dates of
publication.)
Many valuable series of measures have since been published in the " Astrophysical
Journal " and in the " Zeitschrift fur Wissenschaftliche Photographic." The
" International Tables of Constants " also include collections of spectroscopic measure-
ments. Collections of tables for the Schumann region are given in Lyman's book.
CHAPTER II.
HISTORICAL NOTE.
18691879.
The earlier attempts to discover laws governing the distribution of lines in spectra
were controlled mainly by the supposition that the vibrations which give rise to the
lines might be similar to those which occur in the phenomena of sound, and might
correspond with harmonical overtones of a single fundamental vibration. In that
case the ratios of the wave-lengths of different lines would be expected to be repre-
sented by comparatively small integral numbers. Lecoq de Boisbaudran* believed
that he had discovered such relations among the bands of nitrogen, but more exact
measurements which were made later failed to verify his conclusions. In 1871,
however, it was pointed out by Dr. Johnstone Stoneyf that the wave-lengths of the
first, second, and fourth lines of hydrogen were in the inverse ratio of the numbers
20, 27, and 32, and the accuracy of these ratios strongly suggested the existence of
genuine harmonical relations.
The admirable experimental work of Liveing and Dewar,J which extended well
into the ultra-violet part of the spectrum, provided valuable data for further in-
vestigations, and several important features of associated lines were revealed by these
observations. In the spectrum of sodium it was observed that successive pairs of
lines were alternately sharp and diffuse, and that the pairs generally became fainter
and more diffuse as they were more refrangible ; at the same time the distance
between successive pairs was diminished. (Compare Plate I.) It was remarked
that the whole series, excluding the " D " pair, looked very like repetitions of the
same set of vibrations in a harmonic progression, and it seemed that harmonic
relations could be found to subsist between some of the groups. The whole series,
however, could not be represented as simple harmonics of one set of six vibrations
with any degree of probability. Somewhat similar results were also obtained for
potassium, and, later, for the triplets of magnesium. ||
THE SPECTRUM OF HYDROGEN.
The discovery by Huggins^f of a number of prominent lines in the ultra-violet
spectra of Sirius and other white stars (Fig. 1), which seemed to be a con-
tinuation of the regular series of hydrogen lines in the visible spectrum, led
to further search for harmonic ratios in this spectrum on the part of Dr. Johnstone
Stoney.** Evidence that the lines in question were all members of the same physical
system was found in the fact that when their positions were plotted as abscissae
against ordinates which increased uniformly, they fell upon, or very near, a definite
* Comptes Rendus, 69, 694 (1809).
tPhil. Mag., 41, 291 (1871).
J Proc. Roy. Soc., 29, 398 (1879) ; Collected Papers on Spectroscopy, p. 66.
lyiveing and Dewar were careful to explain that their reference to harmonic series of lines
did not imply that the lines were thought to follow the arithmetical law of an ordinary harmonic
progression, but to be comparable with the overtones of a bar or bell.
|| Proc. Roy. Soc., 32, 189 (1881).
jfPhfl. Trans., 171, Pt. II., 669 (1880).
** Quoted by Huggins.
8 Series in Line Spectra. CHAP. n.
curve. A new departure in the investigation was the substitution of the scale of
" wave-frequencies " (the reciprocals of the wave-lengths) for that of wave-lengths.
On forming the first and second differences of these wave-numbers, Dr. Stoney
concluded (erroneously) that the irregularities in the second differences were too great
FIG. 1. THE SPECTRUM OF HYDROGEN : (a) IN SIRIUS, (b) IN VACUUM TUBE.
to be attributed to errors of measurement, and that the lines did not fall exactly on a
smooth curve. Hence it was thought that the lines were not consecutive members
of a single series, but members of two or more series, and attention was drawn to
several apparently harmonic relations between selected groups of lines.
SCHUSTER'S CONCLUSIONS.
A discussion of the evidence for the existence of harmonic ratios in spectra was
given by Schuster in 1881,* and although it was concluded that the number of such-
ratios was not greater than might be attributed to chance, Schuster clearly recognised
that there might be some undiscovered law, which, in special cases, resolved itself
into the law of harmonic ratios. In the light of our present knowledge this is evidently
the case for the ratios of the hydrogen lines discovered by Stoney, which follow
naturally from the simple law which connects all the lines of the series in question ;
the ratios thus have no special significance, but it is interesting to note that they
contain the germ of the true law, and might well have led to its discoveryv
That some law existed was sufficiently evident from the distribution of the lines
in the hydrogen spectrum, and from the maps of the spectra of sodium and potassium
which had been given by Liveing and Dewar, but progress was greatly retarded by
the lack of sufficiently exact measurements of the wave-lengths of the lines. Schuster
found, for instance, that although all the lines of sodium appeared to be double, and
many of those of magnesium triple (compare Plate. II.), the measurements then avail-
able showed no regularity in the distances separating the components: Nevertheless,
ths fact that the hydrogen lines approach each other rapidly as they pass towards
the ultra-violet, and that characteristic groups which are repeated several times in
other spectra also come nearer and nearer together in the more refrangible parts of the
spectrum, was considered by Schuster to furnish a safer basis for further research
than the hypothesis of harmonic ratios. It was, in fact, in this direction that
subsequent advances were made.
* Proc. Roy. Soc., 31, 337 (1881) ; Brit. Assoc. Report (1882), p. 120.
Historical Note. 9
LAW OF CONSTANT SEPARATIONS.
A fact of great importance was established by Hartley in 1883,* namely, that
the components of doublets or triplets occurring in the same spectrum are of identical
separation provided that the positions of the lines are expressed on the scale of
oscillation frequencies (or of reciprocal wave-lengths, which are proportional to the
frequencies) . By the discovery of this law of constant separations, pairs or triplets
occurring in different parts of the same spectrum could be associated with certainty.
A further valuable contribution on the experimental side was made by Liveing
and Dewar in 1883,t when they gave an account of their work on the ultra-violet
spectra of the alkali and alkaline earth metals, and of zinc, thallium, and
aluminium, in each of which there are well-marked series. The characteristics
which they had previously noted in the case of sodium and magnesium were then
found to be equally pronounced in other spectra, and several new series were re-
corded, including what are now called the principal series of lithium, sodium, and
potassium. These observers, however, did not investigate the laws of the series
which they described with such completeness and accuracy.
BALMER'S LAW FOR HYDROGEN.
A new era commenced in 1885, when the law of the hydrogen series was dis-
covered by Balmer.J The number of lines then known to belong to this series, as
produced in the laboratory, had been increased to nine by W. H. Vogel,'and five
more had been recorded by Huggins in the spectra of the white stars. Balmer found
that the series could be represented, probably within the limits of error of the obser-
vations, by a formula of the type
where h is a constant for the series, and mandn are whole numbers. For the actual
lines, using Angstrom's measures of the first four lines, he gave the formula
m z 4
where m takes the values, 3, 4, 5, ....
Thus
Calcd. A Obsd. A 0-C
H a = ?&=6562-08 6562-10 +0-02
5
H 3 = 4 &=4860-80 4860-74 -0-06
a 3
H y = h =4340-00 4340-10 +0-10
ZJL
H & = ?&=4101-30 4101-20 -0-10
8
* Jour. Chem. Soc., 43, 390 (1883).
t iPhil. Trans., 174, 187 (1883) ; Collected Papers, p. 193.
j Wied. Ann., 25, 80 (1885).
Monatsb. Konigl. Acad., Berlin, July 10 (1879).
io Series in Line Spectra. CHAP. n.
The extrapolation to the ultra-violet lines gave values roughly corresponding
to the measured wave-lengths, but it remained rather doubtful whether the law was
exact or only an approximation to the true formula for the entire series. Subse-
quent investigations, however, have shown that the law, with a slightly amended
constant, represents the whole series with extraordinary accuracy.
Although the hydrogen spectrum is in a sense typical of all series, the simple
Balmer formula with modified values of h is not applicable to series in general.
Nearly all attempts to represent series of lines by formulae, however, have been based
upon the Balmer law, with the introduction of one or more correcting terms.
Balmer's discovery of the law of the hydrogen series, together with Hartley's
law of constant separations and Liveing and Dewar's experimental data, provided
a sound basis for further research.
FIRST RESULTS OF KAYSER AND RUNGE.
Shortly after the discovery of Balmer's law, the investigation of series spectra
was taken up by Kayser and Runge, and by Rydberg. The first results of the
former were published in 1888 by Runge,* who announced that formulae had been
found for series of lines of elements other than hydrogen. Their equations were of
the form
, 1 1
/= ; or /=-
where a, b, c are constants special to each series, and m assumes consecutive values
of the series of whole numbers, beginning with 3. The following formula was given
for the principal series of lithium, m being 3 for the line A3232 :
i
A (inmm.) =
T 1 11635r*
It will be observed that the formula is a more general form of that given by
Balmer for hydrogen, which may be written
Kayser and Runge were quick to recognise the need for a more accurate know-
ledge of wave-lengths in such inquiries, and courageously embarked on a new series
of determinations with the aid of a large concave grating, beginning with the spectrum
of iron as a convenient standard of comparison for purposes of interpolation. f
THE WORK OF RYDBERG.
Rydberg made use of data already to hand, and his investigations are of the
utmost importance as having laid the foundation for all subsequent attempts to show
the connection between different series occurring in the same spectrum. His first
* Brit. Assoc. Report (1888), p. 576.
f Abhandl. der Berlin Akad. (1890).
Historical Note. n
memoir was presented to the Swedish Academy of Sciences towards the end of 1889*
and gives a comprehensive account of the results at which he had then arrived. As
in the case of Kayser and Runge, the ultimate purpose of his inquiries .was to
gain a more intimate knowledge of the structure of atoms and molecules, and not-
withstanding the imperfect data then at his disposal, he discovered most of the
important properties of series, and foreshadowed discoveries which were made later,
when experimental work provided the necessary data.
Rydberg commenced his work by sorting out doublets and triplets, largely from
the tables given by Liveing and Dewar and by Hartley, and in this way ascertained
the lines which might properly be associated in the same set of formulae. Although
the terms " doublet " and " triplet " had in general been understood to signify
groups of lines not very far apart, Rydberg showed that there were true doublets
and triplets in which the components were remote from each other, so that while
one component might be in the visible spectrum, another might even be situated
in the ultra-violet, other lines occupying the intermediate spaces. Hartley's law
of constant separations was thus confirmed and extended, with the proviso that the
law was applicable only to members of series of the same species.
Like Johnstone Stoney and Hartley, Rydberg employed the reciprocals of the
wave-lengths in place of the oscillation frequencies themselves, but gave them a
more definite meaning by defining the " wave-number " as the number of wave-
lengths per centimetre ; that is, 10 8 /A in Angstrom Units. He pointed out that the
use of wave-numbers not only saves a great deal of calculation, but is important in
theoretical considerations.
The observation by Liveing and Dewar that pairs or triplets are alternately
sharp and diffuse enabled Rydberg to distinguish two species of series, in addition
to a third species comprising the ultra-violet lines which the same observers had
photographed in the spectra of Li, Na, and K. The first terms of the latter species,
which are the most intense in the spectra, are situated in the visible spectrum and
had not previously been associated with the ultra-violet series. Three chief species
of series were thus recognised as being superposed in the same spectrum, namely :
Principal, including the strongest lines.
Diffuse, of intermediate intensity.
Sharp, including the weakest lines,
and the members of each series might be single, double, or triple.
In each series the distance from line to line diminishes rapidly on passing to
the more refrangible parts of the spectrum, the lines thus converging towards a
definite limit, and in a normal series, intensities also diminish in regular order. In
a graphical representation with the observed wave-lengths (A) or wave-numbers (v)f
of a series as abscissae, and consecutive whole numbers (m) as ordinates, Rydberg
found that each series was represented by a regular curve, which appeared to be
similar in shape for all series, and to approximate to a rectangular hyperbola (see
* Kongl. Svenska Vet.-Akad. Handlingar, Bandet 23, No. 11 (1890) [in French]. Ab-
stracts were given in Comptes Rendus, Feb." (1890); Zeitschr. Phys. Chem., February (1890),
and Phil. Mag., April (1890).
f Rydberg represented wave-number by n, but analogy with the X always employed for
wave-length suggests that v, as used by Ritz, is more appropriate.
12 Series in Line Spectra. CHAP. n.
Figs. 2 and 3, pp. 14 and 17). A rough representation of several series was in
fact obtained by the use of the hyperbolic formula :
C
v=v
where v is the wave-number of a line, m its order number, and v^ , C, and /JL are
constants special to each series ; v x is the limit of the series, being the value of v
when m is infinite.
Further investigation, however, led to the conclusion that the wave-number
should be represented by the equation
v=v w -f(m+jn)
where the form of the function, and any additional constants, would be the same
for all series. The simple hyperbolic formula did not fulfil this imposed condition,
as the value of C was found to vary very considerably from one series to another.
Rydberg then proceeded to investigate the next simplest form of the function, and
adopted it in his subsequent work, namely,
where 2V is constant for all series.
When fj. is zero this formula becomes identical with that of Balmer for hydrogen,
which in terms of wave-numbers, may be written
m 2 4 4v 00 N
~ v m z " v m* ' " Vco ~~ m z
The constant N could thus be calculated from the hydrogen lines, which gave the
value 109,721-6. This, however, was deduced from wave-lengths in air, expressed
on the scale of Angstrom, and was recalculated later as 109,675-00 from wave-
lengths on Rowland's scale, corrected to vacuum.* This number appears in the
formulas for other series, and is generally designated the " Rydberg Constant." Its
value on the international scale has recently been given by W. E. Curtis as
109,678-3.
Rydberg fully recognised that the formula which he employed was only an
approximation to the true function of m or of (m-\-/u,), but by its aid he was able to
trace a large number of series in the spectra of different elements, and to deduce
most of the important properties of series in general.
KAYSER AND RUNGE'S FORMULA.
Almost immediately after the announcement of Rydberg's results, Kayser
and Runge published an account of their investigations of the spectra of the alkali
metals.f These observers also adopted the scale of wave-numbers per centimetre,
and employed the formula
v =A- 2 --
m i m*
where v is the wave-number, m the order number of a line, and A, B, C three
* Congres Internal, de Physique, Paris (1900), p. 211.
f Abhandl. der Berlin Akad. (June 5, 1890).
Historical Note. 13
constants special to each series ; A is evidently the limit of the series ; B is of the
same order of magnitude as Rydberg's constant N, but C varies widely from one
series to another. The formula was by no means capable of giving an accurate
representation of all the lines of a series, and is inferior to that of Rydberg inasmuch
as it fails to show the important connection between the Principal and Sharp series.
A large number of series and many of the properties of series already described by
Rydberg, however, were independently discovered by its use.
Kayser and Runge's formula has passed its period of usefulness, and practically
all the newer formulae represent attempts to improve the original equation of Rydberg.
Kayser and Runge, however, made important contributions to the subject by their
improved tables of the spectra of many elements,* and by their determinations
of the refractive indices of airf which permitted the correction of the wave-numbers
to vacuum.
* Abhandl. der Berlin Akad. (1891, 1892, 1893).
t Abhandl. der Berlin Akad. (1893).
CHAPTER III.
CHARACTERISTICS OF SERIES.
GENERAL FORMULA FOR HYDROGEN.
It should be clearly understood that the series spectrum of hydrogen is of
exceptional simplicity. The Balmer series, however, does not constitute the whole
of the hydrogen spectrum. Another series in the infra-red was predicted by Ritz,
and two members at A18,751 and A12,817-6 have been observed by Paschen. There
is also a series in the Schumann region, of which three members have been
photographed by Lyman. All the series are included in the general formula
v=N ( A m^m
\m^ m i J
where m l =l for Lyman's series, =2 for the Balmer series, and =3 for the Ritz~
Paschen series. For wave-numbers on the international scale, corrected to vacuum,
N=109,678-3 (Curtis).
A graphical representation of these series is given in Fig. 2.
V J10 1OO 9O
70 60
50
44 3O
20
to
a
b
c
TTV
r1
2
8
4-
5
6
7
8
9
m,
1
8
9-\
10
FIG. 2. THE THREE SERIES OF HYDROGEN.
(a) =Lyman series ; (b) =B aimer series ; (c) =Ritz-Paschen series.
THE FOUR CHIEF SERIES OF OTHER ELEMENTS.
The work of Rydberg, and that of Kayser and Runge, revealed the existence
of a large number of series, each of which is generally similar to the Balmer series
of hydrogen. In each series the lines become closer, with diminishing intensities,
in passing from the red towards the violet end of the spectrum, and converge to a
Characteristics of Series. 15
definite limit (Fig. 3, p. 17). Theoretically, the number of lines in a series is infinite,
but no series have actually been traced to their limits. There are no known series in
which the lines converge towards the less-refrangible part of the spectrum.
In the general case there are several series superposed in the same spectrum.
Several such series are intimately related to each other and may be conveniently
regarded as forming a " system " of series. Four of the related series have a certain
amount of independence and may be considered to be the chief series of a system.
The remainder may be looked upon as derived series.
Three of the chief series were recognised by Rydberg and by Kayser and Runge,
namely, in order of intensity :
Principal, Diffuse, Sharp (Rydberg).
Principal, first Subordinate, second Subordinate (Kayser and Runge).
Rydberg's names are most commonly used and are conveniently abbreviated to
P, D, S, while the respective limits may be represented by Poo , Z)oo , Soo .
The fourth chief series long escaped detection because many of them occur in
the infra-red, in which region they were first observed by Bergmann.* They are
called " Bergmann series " by some writers, but as they do not all occur in the
infra-red, and were not all discovered by Bergmann, the name is not specially
appropriate. From theoretical considerations Hicks has named them the
" Fundamental " or " F " series, and though they are probably not more funda-
mental than the other chief series, the name has been so much employed that it will
be convenient to retain it for the purposes of the present report.
Although less exact than some of the amended forms which have since been
proposed, Rydberg's formula is usually a sufficient approximation to bring out the
main characteristics of a system of series, and its simplicity is a great advantage in
approaching the subject. This formula will accordingly be adopted for descriptive
purposes, namely :
N
where A is the limit of the series, N the Rydberg constant for hydrogen, and the
wave-numbers v m are; obtained by assigning successive integral values to m ; p may
be regarded as a decimal part to m, though it is sometimes greater than unity.
Each series is thus represented by a limit, and a series of " variable parts " or
" terms " forming a " sequence."
The four chief series may be represented by the Rydberg formulae :
Principal ......... P(w)=Poo N/(m+P) 2 m=l, 2, 3 . . .
Sharp ......... S (w)=Soo -iV/(w+S) 2 m=2, 3, 4 . . .
Diffuse ......... D(m)=Doo -N'/(m+D) z w=2, 3, 4 . . .
Fundamental ...... F(m)=Foo NI(m+F) z m=3, 4, 5 . . .
where P(m), for example, means the w-th line of the P series, and P, S, D, F
indicate the values of p in the respective series.
These formulae represent the four chief series of a " singlet " system, but in
many systems each member of a series is a doublet or a triplet.
* Dissertation, Jena (1907); Zeit. Wiss. Phot., 6, 113, 145 (1908).
16 Series in Line Spectra. CHAP. HI.
RELATIONS BETWEEN THE CHIEF SERIES.
The relations between the P, S, and D series were most completely traced by
Rydberg. His method may be usefully illustrated by reference to the spectrum
of lithium, which consists of doublets close enough to be regarded as single lines.
Rydberg's original formulae, which are sufficiently accurate for our purpose, were
as follows :
x 109,721-6
w+0-5951) 2
109,721-6
5(w) =28,601-1 -
Dim) =28,598-5--
(w+0-9974) 2
These formulae, and similar ones calculated for series of other elements, showed,
in the first place, that the limits of the 5 and D series were probably identical, i.e.,
Sao =Dco .
Next, it was found that m=l in the variable part of the formula for S(m) gave
approximately the limit of the P series ; thus
W/(l-5951) 2 =43,123-7
And, similarly, ml in the formula for P(m] gave approximately the limit of
the S series ; thus
N/(l-9596) 2 =28,573-l
From such considerations Rydberg concluded that Poo =JV/(1+S) 2 and
Soo =Doo =N/(l-\-P) z , so that the three series could be represented by
N N
N N
(m+D) z
THE RYDBERG-SCHUSTER LAW.
The first line of the P series (given by m=l) is thus identical with that given
by m = l in the S formula, but with opposite sign: i.e., S{1) = P(l). It follows
that the difference between the limit of the P series and the^ common limit of the D and
S series is equal to the wave-number of the first line of the P series. This important
rule was clearly included in the formulae given above, but Rydberg did not express it
in these terms until 1896,* in which year Schusterf also independently announced
its discovery, apparently by reference to the limits of the numerous series which
had been calculated by Kayser and Runge. The law of limits is thus generally
known as the Rydberg-Schuster law, and is expressed symbolically by
^oo -$ - P (1)
* Astrophys. Jour., 4, 91 (August, 1896).
f Nature, 55, 196, 200, 223 (1896).
Characteristics of Series.
RUNGE'S LAW.
Runge* was the first to point out that the difference between the limits of the
D and F series is equal to the wave-number of the first D line (mostly given by
m=2 in Rydberg's formulae). This relation is often referred to as Runge's law,
and is expressed symbolically by
#00 -Foe * D (2)
Thus, in Rydberg's form,
In all series of this type the value of F approximates to unity,
ABBREVIATED NOTATION.
A convenient and now indispensable abbreviated notation for series was suggested
by Ritz in connection with the more complex formula which he employed. It is,
however, of general application, and merely provides that a term NI(m-\-/j,) z or its
V45
30
25
20
o
s
D
F
Tit
1
2
-S
4-
6
6
7
8
9
10
1 -
2
3
4>-
5-
6-
7-
8-
9-
10-
Foo
FIG. 3. THE FOUR CHIEF SERIES OP LITHIUM.
equivalent should be represented by nifi, or by mP, mS, mD, mF in relation to
individual series. Thus the four chief series are written
P(m) = ISmP
S(m) = IPmS
D(m) = IPmD
F(m) = 2DmF
It is to be understood that while .0(3), for example, indicates the line in the D serie.8
for which m=3, 3D is equivalent to the term N/ (3 +Z)) 2 'and represents the interval
from the line D(3) to D x .
* Phys. Zeit-. 9, 1 (1908).
i8 Series in Line Spectra. CHAP. m.
The foregoing relations, in the case of lithium, are represented graphically
in Fig. 3, p. 17.
In order to distinguish series of different kinds, Paschen designated singlet
systems by the use of capital letters, and doublets and triplets by small letters.
Following a suggestion made by Prof. Saunders, however, it will be convenient to
adopt Greek letters for doublets, and small letters for triplets. Thus
P, S, D, F = Singlet systems.
it, CT, d, 9 = Doublet ,,
p, s, d, f = Triplet
For general descriptive purposes, however, we shall occasionally use P, S, D, F
for any class of series.
DOUBLET SYSTEMS.
The special characteristics of a doublet system, such as that of sodium (see
Plate I.), assuming that there are no complications due to satellites, may be
briefly stated as follows :
(1) In the d and cr series, the less refrangible components of the pairs are the
stronger, and the separations of the components, when expressed in
wave-numbers, are constant throughout.
(2) In the n series, the more refrangible components of the pairs are the
stronger, and the first pair has the same separation as d and er. The
components approach each other as the order number increases, and the
two series have the same limit.
(3) The 9 series consists of single lines.
These characteristics are embodied in the following formulae, where the brighter
components are indicated by n v a v d v and the fainter by n 2 , a z , 6 2 :
Rydberg formulae. Abbreviations.
(jii(m)=NI(l+a) z Nl(m+nf = ICT mn^ Shorter A
\7r 2 (w)=Ar/(l+ar) 2 JV/(w+;r 2 ) 2 = l
^"^
_C
P-i
CL,
Characteristics of Series. 23
Doublet series with positive and negative P(l) are thus illustrated in Fig. 5.
When P(l) occurs with negative sign in a triplet series, as is usually the case, the
first triplet is inverted, the brightest component and the wider separation being then
on the less refrangible side, as in the triplets of the sharp and diffuse series. In the
remaining triplets, however, the stronger component and wider separation are on the
more refrangible side.
The first member of the diffuse series also sometimes appears with negative sign,
and the corresponding doublet or triplet is then reversed right and left in the actual
spectrum. An interesting example appears in Fig. 6, where the lines 8662, 8542,
8498 of ionised calcium (see p. 25) form an inverted diffuse doublet with satellite.
In the tables which accompany this report negative members are
indicated by the usual minus sign. The occurrence of these negative signs is probably
of no great theoretical importance, because the " terms " from which the lines are
derived by taking differences are always positive.
THE COMBINATION PRINCIPLE.
The possible existence of other series which would be related to the chief series
already considered was first suggested by Rydberg,* who pointed out that in his
general formula
L 1
where m was usually =1, it might be supposed very probable that wtjas well
would be variable. Such variation oi m 1 would evidently give rise to other series
running parallel to the first. Rydberg was unable to establish any such series, but
the idea was developed later by Ritz.f
The lines of a series always appear as the difference of two terms, one of which
is the limit of the series, and the other a " variable part " or " term " of a sequence
given by successive integral values of m ; the limit itself in the case of the chief series
is the first term of the sequence of one of the other series. Ritz discovered that lines
which were computed by taking differences of other terms of the four cliief sequences
were often found in the actual spectra, and he called them " combination lines," or
" combination series " when more than one line was derived from the same sequence.
Thus, there might be series 2 5 mP, 3 S mP, as well as the chief series, 1 SmP ;
or 1 SmD, 1 PmP, and so on. The fundamental, or " Bergmann," series was at
first regarded by Ritz as the combination series, 2D wAP, as already explained on
p. 18, but is now to be considered as one of the chief series.
The extensive investigations of infra-red spectra subsequently made by Paschen
and his pupils have revealed a very large number of combination lines, and have
shown that the combination principle is probably exact. The application of this
principle has shown that many lines which previously appeared to be unattached
really form part of regular systems, of which the principal, sharp, diffuse and funda-
mental series are the chief members, and it seems possible that in spectra in which
series have not yet been identified the complexity may be due to the existence of a
great number of combinations.
' It is interesting to observe that the proof of the combination principle is indepen-
dent of a knowledge of the true series formula, or even of the exact limits of the various
t
Loc. cit., p. 73.
Phys. Zeit., 9, 521 (1908) ; Astrophys. Jour., 28, 237 (1908).
24 Series in Line Spectra. CHAP. in.
series. The limits can, in fact, be determined with considerable accuracy in most
cases ; but by assuming the Rydberg and Runge laws to be exact, and adopting a
limit for one of the series, all the limits will be equally in error. Then, if the variable
terms mp, ms, &c., be derived by subtracting the observed wave-numbers of the lines
from the adopted limits of their respective series, these terms will be equally affected
by any error in the limit first adopted. Hence, the errors will cancel each other in
taking differences to form the various combinations. An example may be taken from
lithium (see table for this element) by way of illustration. The limit calculated for
the principal series is 43,486-3, and n (!) is 14,903-8 ; thus
n^ .=43,486-3=1 , and from the relation between the a and n series we have
Av=cjoo -(joo ^
where n z and n may include extra terms introduced for the better representation
of the series. In the Ritz formulae, the denominator terms are indicated by
(m -\-pi~\-n' I m 2 ) and (m-\-p z -}-7i"/m z ), and Ritz found that the quantities (Pi~P^
gave a much smaller range of values in relation to the squares of the atomic weights
than did the separations themselves when different groups of elements were included
in the comparison. His results are shown in the following table :
Na
K
Rb
Cs
Cu
Ag
Mg
Ca
Sr
Zn
Cd
Hg
Al
In
Tl
He
A^.IO'/W"
32-3
37-8
32-2
31-6
61-8
79-0
68-8
66-1
51-5
91-0
93-2
115-4
152-8
172-1
187-0
63-8
in 7
14-2
18-9
18-0
18-6
18-6
24-2
14-6
17-7
15-7
17-2
18-6
22-3
24-8
29-2
32-7
20-4
* Zeit. Phys. Chem., 50, 100 (1904).
t Phil. Mag., 18, 411 (1909).
t. Proc. Nat. Acad. Sci., Wash., 3, 409 (1917).
Astrophys. Jour., 28, 241 (1908) ; 29, 243 (1909).
Spectra and Atomic Constants. 47
The essential difference of procedure as compared with the use of direct values of
Av is that in the form (pj_p 2 ) the doublet or triplet separations are associated with
the limits of the subordinate series. Ritz regarded these results as merely preliminary,
but does not appear to have carried the matter further. A more exhaustive study,
however, has been made by Hicks (see Chapter VIII.).
The results obtained by substituting atomic volumes for atomic weights in
relation to the doublet and triplet separations in the foregoing comparisons do not
appear to be any more definite than those already quoted. Thus, in the elements
of the alkali group, potassium still deviates from the curve which connects atomic
volumes with the separations, or from the straight line when the logarithms are
plotted.
HOMOLOGOUS LINES AND ATOMIC WEIGHTS.
In an attempt to discover the relations, between the spectra and other physical
properties of the elements, Ramage* compared graphically the corresponding or
" homologous " lines in the different elements of the various groups. The character-
istic flame lines Mg 2852, Ca 4227, Sr 4607, Ba 5535, are examples of such lines ;
and, in general, lines of corresponding series which have the same value of m would
be regarded as homologous. Such lines, like the limits of the series to which they
belong, usually advance towards the red with increase of atomic weight. The curves
connecting their wave-numbers with the atomic weights, or with the squares of the
atomic weights, however, were found by Ramage not to be continuous throughout
the whole of a group of elements. In the alkalies, for instance, there was a break
between Na and K. It does not seem probable that the discussion of individual
lines will lead to more definite relations than those of the limits of the series.
Ramage also derived empirical formulae for the series of K, Rb and Cs, in which
the atomic weight was the only variable, apart from the parameter m. Thus, he
found that the second components of the principal series were given by
v=35,349-0-2233J7 2 -iV/[w+M9126+0-00103JF
+(0-04377 +13W* x 10- 7 ) (1 -S 1 '" 1 )] 2 .
The possibility of obtaining such a formula strongly suggested to Ramage that
the diff ere ices between the series of the three elements in question depended solely
upon differences of atomic weights. The number of constants involved, however,
is too large to give confidence in such a conclusion, especially as the formula fails
to include lithium and sodium.
Other relationships between homologous lines and atomic weights were afterwards
suggested by Marshall Watts.f Thus, in each of the two groups K, Rb, Cs, and Ca,
Sr, Ba, the differences of wave numbers between corresponding lines in their spectra
were found to be nearly proportional to the differences between the squares of the
atomic weights. In the first group the lines having the following wave-numbers
were assumed to be homologous :
Cs Rb K
(a) 12,469 (6) ... 13,742 (4) ... 14,465 (7)
(b) 21,764 (6) ... 23,714 (6) ... 24,700 (6)
(c) 21,945 (8) ... 23,791 (8) ... - 24,719 (8)
&c. &c. &c.
* Proc. Roy. Soc., 70, 1 and 303 (1901-02).
t Phil. Mag., (6) 5, 203 (1902).
48 Series in Line Spectra. CHAP. vn.
Adopting atomic weights 132-7 for Cs and 39-9 for K, the rule gives with lines (a)
86-87, with lines (b) 83-24, and with lines (c) 83-11 for the atomic weight of Rb (85-2).
The limits of the principal series, treated in the same manner, give 86-0. Reference to
the general tables will show that the lines (a) are not homologous, and if the correct line
of Cs(13,138) be introduced the deduced atomic weight for Rb becomes 102. The
suggested relation also excludes lithium and sodium and would therefore seem to
have no great significance.
In another class of elements, represented by Zn, Cd, Hg and Ga, In, Watts found
that the differences between the wave-numbers of certain lines of one element
were to the differences between the corresponding lines of the other elements as the
squares of their atomic weights. Among the pairs of assumed corresponding lines
of Cd and Zn given by Watts are the following :
Cd Zn
(a) 30654-4 (10) ..." ............... 32500-0 (8)
(b) 30734-9 (8) .................. 32540-1 (10)
(c) 31905-5 (8) .... .............. 32928-7 (10)
(d) 32446-8 (6) .................. 33118-6 (8)
&c. &c.
The assumption of 111-83 for the atomic weight of Cd then gives 65-44 from (a)
and (c), 65-69 from (a) and (d), and so on, for that of Zn (64'9). Since lines (b), (c) and (d)
belong to corresponding triplets, and lines (a] are not placed in series, it will be seen
that in place of taking the triplet separations themselves to be proportional to the
squares of the atomic weights, Watts has added wave-numbers to the separations
equal to the intervals between the lines (a) and (b) of the two elements. That is,
since (c) (b)=&v lt and (d) (c)=Av 2 , if (b) (a) be called x for Cd and y for Zn.
Watts makes
x (Cd)
For (a) and (c)
jy+Av/ (Zn)
and for (a) and (d)
(Zn) 2
The difference between the two results from members of undoubtedly corre-
sponding triplets, measured with sufficient accuracy in each case, sufficiently indi-
cates that this mode of correcting the triplet separations in forming the ratio of
squares of atomic weights cannot be valid. In fact, the two expressions on the left
could only be equal if Av 2 /Av a * were also equal (Cd) 2 /(Zn) 2 , and this is only approxi-
mately true for these two elements, and very far from true when mercury is taken
as one of the elements. Some of the other lines taken to be homologous by Watts
are certainly not corresponding lines, and it is difficult to understand on what prin-
ciple they were selected. The deduced values for Zn range from 64-77 to 67-08.
SEPARATIONS AND ATOMIC NUMBERS.
Since atomic numbers probably determine the places of the elements in the
periodic table more correctly than the atomic weights, several attempts have been
made to correlate these numbers with the series spectra.
In a graphical repetition of Runge and Precht's work on the doublet separations
Spectra and Atomic Constants. 49
in the calcium group Ives and Stuhlmann* found that the results were somewhat
more consistent than for the atomic weights, but the atomic number derived for
radium was 96, in place of the true value 88. Using the wave-number differences
between extreme members of the triplets occurring in elements of the same group,
it was found also by Anslow and Howell| that when the logarithms of these differences
were plotted against the logarithms of the atomic numbers, the points fell more
accurately on a straight line than with atomic weights, and an atomic number of
87 was deduced for radium.
Separations in relation to atomic numbers have also been discussed by tL
Bell, | who employed two formulae, namely :
(1) A /A7=m(N-.V ) ; (2) log Av=/> log N+q
where AV is the separation, N the atomic number, and the other terms are constants
to be calculated for each group of elements. The figures for the alkali group will
serve for illustration :
At. No. Aj/ obs. Ay calc. (1). Ai/ calc. (2).
L,i 3 ... 0-34, ... 0-25 .:. 1-03
Na 11 ... 17-21 ... 16-48 ... 17-21
K 19 ... 57-90 ... 58-0 ... 56-17
Rb 37 ... 237-71 ... 244-0 ... 237-7
Cs 55 ... 564-10 ... 558-3 ... 560-8
m =0-4447; ^0 = 1-875 ; = 2-1645 ; ? = 1-01832.
A somewhat similar formula was tested by Paulson, || namely
log &=A log (N+n)+B
where N is the atomic number, Av the separation, A and B constants, and n a
positive or negative integer. For each group of elements the value of n was first
determined graphically, and the constants A and B were then calculated ty the
method of least squares. The atomic numbers used were those of Rydberg's system ,^|
which are two units higher than those of Moseley. The nature of Paulson's results
may be gathered from the following data for the triplets of the calcium group :
log Av 1 =2-163129 log N -0-871542
log Av 2 =l-748748 log (N -4) -0-459734
Mg
Ca
Sr
Ba
Ra
The extrapolation to radium does not agree at all closely with the separations
2016-64 and 1036-15 afterwards suggested by Anslow and Howell. It will be observed
* Phys. Rev., 5, 368 (1915).
"i* T oc cit
i Phil. Mag., (6) 36, 337 (1918).
J; Not used in calculation of constants.
ij Astrophys. Jour., 49, 276 (1919).
If Jour. Ch Phys.. 12, No. 5 (1914).
N.
(Rydberg.)
A V1 .
Ai/ 2 .
Ai/! calc.
C.
Ai/2 calc.
C.
14
40-92
19-89
40-52
+0-40
19-45
-j-0-44
22
105-99
52-11
107-72
1-73
54-38
-2-27
40
394-44
187-05
392-57
+ 1-87
182-74
-14-31
58
878-4
370-3
876-96
+ 1-44
371-35
1-05
90
(2268)
(838)
5 C Series in Line Spectra. CHAP. vn.
that no simple connection between the atomic numbers and the doublet or triplet
separations has yet been discovered.
There does not appear to be any published record of investigations of the limits
of series in relation to atomic numbers, but a few trials will show that the results
are generally similar, and not more exact, than for the atomic weights. Thus, in
the alkali group, potassium remains decidedly discordant, and in the calcium group,
calcium again shows considerable departure from the approximate regularity shown
by the other four elements. When the limits of the principal series are plotted against
atomic numbers the points show the same absence of simple regularity which was
found with the atomic weights.
CONCLUSIONS.
These results are in a sense disappointing. It would seem that the spectra
'must for the present be regarded as constants of the elements which show no simpler
relation to other constants than is shown by some of the constants among them-
selves. Thus, in the alkali group, the curve connecting atomic weights with melting
points, or that connecting atomic weights and atomic volumes, is closely similar to that
relating the limits of the subordinate series to atomic weights, and a similar discrep-
ancy is shown by potassium in each case. Again, there is no simple relation between
the atomic weights and densities in this group of elements, just as there is no simple
law connecting the limits of the principal series. It can only be concluded that
although the spectra change progressively with atomic weights, atomic volumes or
atomic numbers, the laws governing the changes are not clearly indicated by any
of the foregoing invest : gations.
CHAPTER VIII.
THE WORK OF HICKS.
The discussion of spectral series which has been made by Prof. W. M. Hicks* is
of so special a character that it is most conveniently treated separately. The
investigation covers a great deal of ground, and it will only be possible to attempt to
give a general idea of some of the methods employed. In the earlier papers Hicks
proves the value of the series formula
N
= A-
/ . . _\
V m)
and makes a special feature of calculating the influence of possible errors of observa-
tion on the values of the constants for the different series. Having determined the
constants for many of the known series, he proceeds to discuss them in relation to the
atomic volumes and atomic weights of the respective elements. The adopted wave-
lengths are on Rowland's scale, and as the conclusions would not be modified by the
substitution of wave-lengths on the international scale, it has not been thought
necessary to recompute the formula constants and the quantities which depend upon
them.
It should be remarked that in some of his papers Hicks departs from the more
usual practice, and writes, for example, p(l), instead of 1 p, for the variable part
or " term " of the first principal line.
ATOMIC VOLUMES.
Among the more striking results obtained by Hicks in his first paper is the follow-
ing comparison of the constants for the stronger components of the principal series
of the alkali metals with the atomic volumes. The limits of the series are also entered
for completeness :
<"
(X,
** 1
H 1+a
At. vol.
Ivimit.
u ...
0-951609
+0-007365
1X11-81
43,486
Na...
1-148678
0-031776
2X0-074339
2X0-058451
2X11-80
41,449
K ...
1-296480
0-062511
4X0-074120
4X0-058492
4X11-15
35,006
Rb. M
1-366399
0-074554
5X0-073280
5X0-058369
5X11-21
33,689
Cs ...
1-450967
0-090077
6X0-075161
6X0-060148
6X11-76
31,405
The constants (jn 1) and (ft 1+a) are thus approximately integral multiples
of constant numbers, and the atomic volumes the same multiples of another number.
Lithium is excluded in this mode of treating the formula constants, and Hicks has
suggested that what is usually regarded as the principal series of lithium may really
be a combination Ismd or Ismf; the easy reversal of the lines, and the similarity
to the principal series of the other alkali elements, however, is directly opposed to
this supposition. Assuming the above relation to be sufficiently general, and
observing that a/( / a 1) is nearly constant, Hicks endeavoured to bring the figures
into closer agreement by the introduction of an atomic weight term as a correction to
* I.. Phil. Trans., A. 210, 57-111 (1910); II., 212, 33-73 (1912) ; III.. 213, 323-420 (1914) ;
IV., 217. 361-410 (1917) ; V.,220, 335-468 (1919). Corresponding numbers are used for references
to these papers in the text.
E 2
Series in Line Spectra.
CHAP. VIII.
ATOMIC WEIGHT TERM.
The atomic weight term, as defined by Hicks, depends jointly upon the separa-
tions of doublets or triplets, and the limits of the subordinate series. Thus, if D
represents denominator, the formulae for the two members of a sharp series of doublets
may be written :
A is then the atomic weight term as represented in the later papers of Hicks, and is
equivalent to the 2W of the first paper. In triplet series the two separations give
A! and A 2 , of which the former is the greater.
Hicks was led to conclude that there is a universal constant, approximately
0-21520, so that the denominator of the principal sequences of the alkalies, or of the
sharp sequences in the elements of the second and third groups might be represented by
m/
m
where a and k are constants for all elements, not very different from 0-002740/(1 k)
and 0-21520 respectively, v is the atomic volume, and s is an integer special to each
element as follows :
Na 2
K 2
Mg 8
Ca 7
Zn 9
Cd 8
Al 8
Ga 8
Rb 2
Cs 2
Sr 6
Hg 6
In 6
Tl 5
If k be taken =0-21520, a =0-003490. W is to be taken as A/2 for doublets, and
as A 2 for triplets.
In order to indicate the degree of approximation given by the general formula,
the following comparison may be made with the values of the denominators calculated
from the regular series formulae :
Element.
Series.
At. vol.
s
A/2
A.
True
denominator.
From general
formula.
K ...
KI
44-60
2
0-001466
(1) 2-233969
2-244315
(2) 3-265225
3-277078
Ca ...
s i
25-5
7
0-001368
(1) 2-484198
2-488903
(2) 3-522799
3-555250
Zn ...
Sl
9-33
9
0-003475
(1) 2-227899
2-229991
(2) 3-257479
3-259785
It will be seen that the correspondence is far from perfect, and after the exhaustive
investigation made by Hicks it may be doubted whether it is really possible to deduce
such a formula which shall be applicable to all elements. The atomic volume,
given by atomic weight divided by density, as Hicks remarks, varies with temperature
and cannot be directly involved in the spectral relations ; it would seem, however,,
that it may be closely related to an atomic property analogous to volume, or sphere
of action, on which the structure of the spectrum in part depends.
It should be particularly noted that the atomic volume term, according to Hicks,
does not appear in connection with the diffuse and fundamental series ; in the
alkalies it is associated with the principal sequence, and in other groups of elements,
with the sharp sequence.
The Work of Hicks.
53
THE OUN.
Reference has already been made (p. 46) to the discussion by Ritz of the
terms {p 1 p 2 )/W 2 , from which it appeared that the range of values was much less
for all elements than in the case of &v/W 2 . Hicks (III.) has made a somewhat similar
investigation, but has taken the whole denominator terms into account in forming
the ratios. Taking W as the atomic weight of an element, wW/100, A the
denominator difference which gives rise to the doublet separations, A x and A 2 the
corresponding numbers for triplets, Hicks concludes that A is in all cases a multiple
of qw 2 , where q has the same value for all elements (=about 90-5, when A is multiplied
by 10 6 ). The quantity qw 2 is that which Hicks has named the oun (cov), each element
thus having an oun peculiar to itself and dependent upon the atomic weight. The
oun is designated d v but the multiple 4<5j occurs most frequently and is indicated
by <5. We thus have :
6 1 =qw 2 =l oun for element of atomic weight 100o>
<5=4g7> 2 =4 ouns
10 6 A=W(3 1 or md, where m is an integer special to each element.
The derivation of the calcium oun may be taken for illustration. The triplet
separations are 105-89 and 52-09. The limits of the sharp series are 33983-45,
34089-34, and 34141-43, and when these are put in the form N/D 2 , the denominators
(D) are 1-796470, 1-793679, 1-792310. The denominator differences are therefore
A 1 =0-002791, A 2 =0-001369, which are multiplied by 10 6 and tabulated as 2,791
and 1,369. The atomic weight is taken to be 40-124, so that 10 6 A 1 /o> 2 =17336-l=48
X 361-169, where the last factor is the number nearest to 4<7 which makes the multi-
plier an integer. The value of 6 for calcium is thus 361-169 x(0-40124) 2 =58-14 ;
or <3=10 6 A 1 /48=58-14.
This process has been applied by Hicks to a large number of series, and some of
the results are given in the following table :
(jjoo
W=100w
AV
(Sepn.)
10 6 A
10 6 A/ze>*
mq
or m.&q
8=4?zw s
= 4 ouns
361-8
Na
K
Rb
Cs
24475-40
21964-44
20871-29
19673-00
22-998
39-097
85-448
132-823
17-175
57-87
237-54
553-80
743
2,939
12,935
32,551
14027-96
19224-86
17715-86
18449-48
155X 90-50
53X 362-72
49X361-40
51X361-74
19-17
55-45
263-77
638-22
0-2, 0-14
0-92, 3-22
0-40, 0-56
0-06, 0-33
The last column gives the difference between 361-8 and the number under 4?,
except that when it is not the 4 x 90-5 term it is brought up to it by multiplying by 4.
The second set of figures in this column shows the limits of permissible variation of
the deduced 4^, due to uncertainties in the measures of the lines. In nearly all cases
the differences from 361-8 are within the possible errors depending upon imperfect data.
In this way, Hicks found from 17 elements, weighted according to the possible
errors
1 oun=<9 1 =(90-47250-013)w 2
4 ouns=<5 =361 -89w 2
The oun, according to Hicks, appears in connection with series in several other
ways. The satellite separations, for example, are dependent upon multiples of the
oun, and in the case of triplets A x 2A 2 is also an oun multiple.
54
Series in Line Spectra.
CHAP. VIII.
It should be observed that formulae for series do not enter into this discussion,
except as regards the adopted value of N and the determination of limits. Except
when the oun is very small, Hicks believes that with the accuracy now attainable in
spectroscopic observations, it should be possible to obtain far more reliable values
of the ouns of the various elements, and thence of the atomic weights, than by pro-
cesses depending upon weighing. There is, however, usually no apparent regularity
in the multiplying integers which determine the doublet or triplet separations, even
among elements of the same group, so that the simplest application of the spectro-
scopic method is to assume an approximate atomic weight in order to evaluate the
multiple, and then to use this to correct the atomic weight.
By this method Hicks has since made determinations of the atomic weights
of copper and gold, taking silver as standard with atomic weight 107-88.* The
results may be summarised as follows :
Doublet
separation
(A,)
Sioo
10 6 A
S
Multiple
of 8
W = 100 ^8/4?
Cu
248-44
31523-48
7307-087
146-1419
50
63-5569 -006
AK .
920-438
30644-60
27786-57
421-047
66
107-88(assumed)
Au
3815-56
29469-85
113951-00
1406-802
81
197-193-003
The value of the constant 4? deduced from Ag is 361 -7837 -0038, and this
was used in calculating the atomic weights of Cu and Au from the d terms. Brauner's
values are 63-56^-01, 197-20-07 respectively for these elements. It should be
observed that the separations of the pairs of lines involved in these determinations
can be measured with great accuracy, but that the exact determination of the limits
of the series is less certain.
In some cases the spectroscopic determination may be made independently
of any previous knowledge of the atomic weight, by utilising the various different
ways in which the oun is considered to play a part in building up the spectrum,
and finding the smallest common factor. An illustration is given later (p. 58).
COLLATERALS.
Further evidence of the oun as a controlling influence on the spectrum is adduced
by Hicks from the supposed existence of what he has called " collaterals." In the
case of doublets and triplets, the second, or second and third components may be
considered as having received a sort of lateral displacement by the atomic weight
term A, or Aj and A 2 , and may thus be regarded as collaterals of the first. Hicks
believes that this kind of displacement is not restricted to doublets and triplets,
or their satellites, but is of very common occurrence. Thus, if the wave-number
of a series line be N/D-fN/D^, lateral displacements may be produced by the
addition or subtraction of multiples of d or A, say xd or #A, to D or D m . When
added to D 1 the operation is indicated by writing xd to the left of the symbol of the
original line, and when added to D m to the right. As an illustration, Hicks
takes the Ca line at 1 6439-36. This may be represented by (2A 1 +10A 2 )
Ca s 1 (2)( + A 2 ), meaning that whereas
Wave No. of Cas,(2) = -
1V ' (1-796470) 2 (2-484994) 2
* Phil. Mag., 38, 6 & 301 ; 39, 457.
Wave No. of Ca 6439-36=
The Work of Hicks. 55
N N
;i-79647U+2A 1 +10A 2 ) 2 (2484994 + A 2 ) 2
N N
(1-815732) 2 (2-4S6362) 2
2791 ; (10 6 )A 2 =1369.
In this way Hicks accounts for irregularities in the satellite separations which
sometimes occur, and for discrepancies between the observed positions of certain
lines and the positions calculated from formulae. In some cases he considers that
the whole set of lines for a given order number m may be replaced by another
strong set displaced by several multiples of A, or by a congeries of fainter lines
displaced by various oun multiples.
The whole procedure, however, seems to be somewhat arbitrary, and it remains
to be seen how it will bear the test when observations of sufficient accuracy for such a
purpose become available. Some of the examples first mentioned by Hicks are
certainly no longer admissible; Mgdj(4), for instance (III., 356), has since been
shown to be perfectly normal by the resolution of the line into two components.*
CONSTITUTION OF DIFFUSE SERIES.
Hicks has further concluded that the diffuse and fundamental series cannot be
represented by a continuous mathematical expression, though they may approximate
to values so represented. He considers it more probable that they depend on
discrete changes which are connected with the oun, or atomic weight term, in a
way which has yet to be discovered. Thus, successive denominators (D) of a diffuse
sequence are thought to differ by integral multiples of the oun. When there are no
satellites, the denominators change by multiples of A in the case of doublets, and of
A 2 in the case of triplets, except that in the oxygen group the multiples are of
Aj. . When satellites are present, the multiples are of 6 or d v In addition, the
decimal part, or mantissa, of the denominator of the normal first line of the diffuse
series is itself a multiple of A, the outer satellite being taken as the normal line.
The general character of this part of the investigation will be sufficiently indicated
by the first chief line and satellite of the diffuse series of caesium :
Cs (10 6 )A=32551, (10 6 )<5 =638-22, Doo =19673-0.
Chief line. Satellite.
2-554329(228) -76|-43 46^ 2-546989(226) -97
30(5
3-535183(200) -2011+40 54^ 3-526567(200) +9
10^
4-533588(160) 424|+1 146 4-524635-161
3<5j
5-533110(400) -768^+22 14<5 5-524175-26
&c.
546989=857(638-2600-233 0-08871) =857<5
In this table the denominator terms have been adjusted within permissible
limits indicated by the possible errors of observation. The number in brackets
following the denominator term is the estimated limit of error in the last three
digits, | is the error of the limit, and the last number represents the difference
* Fowler & Reynolds, Proc. Roy. Soc., A. 89. 139 (1913).
56 Series in Line Spectra. CHAP. vm.
between the observed value of the denominator and the " selected " value entered
in the table. The multiples of 6 or 6 1 in the middle of the table are the satellite
separations, expressed in terms of denominator differences.
It is extremely difficult to form a just estimate of the confidence which may be
placed in these results, on account of uncertainties in many of the observational
data, and the occasional exceptions to the more general rules above stated. Thus
Cs is exceptional inasmuch as the first mantissa is a multiple of 6 and not of A ;
and Cd is irregular because it is the mantissa of the chief line, and not that of the
outer satellite, which is a multiple of A. Very few spectra have been measured
with the accuracy and completeness which would seem to be necessary to justify
the deductions fully, and the adjustment of data within estimated permissible ranges
is not an entirely satisfactory substitute. The apparent absence of any general
law governing the sequence of multiplying integers in the successive denominator
terms is somewhat disappointing, for it is clear that the oun theory does not yet
provide a guide to the identification of series lines such as is provided even by an
approximate formula.
LINKS.
In his fourth paper, Hicks has extended the idea of collateral displacements
with a view to associating the lines which do not fall into the ordinary series with
those which belong to the regular series systems. Each line of a series is regarded
as being connected with other lines in the same spectrum through several constant
differences of wave-number, or links, which may be added or subtracted to an in-
definite extent, and apparently in any order. The various links which occur in
a doublet system are distinguished by letters which have the following meanings,
as given by Hicks, and as written in the more extended notation of this report :
>!- A)
e=p(-3 A) -p(&) =NI(l +#!-3 A) 2 -AT/(1 +^+ A) 2
w=s-s(A)
As before, A is the denominator difference corresponding with the normal
doublet separation. The link b is the normal doublet separation, and link e=
a+b+c+d.
These methods were first applied to the spark spectra of silver and gold, con-
taining 600 and 741 lines respectively in the region covered by the investigation.
It will suffice to take silver as an illustration :
Ag. Links. Corrections to links.
^>=<7 l0 o =3064,4-60 a=880-77 a' a = 0-61*
=A/7(1-891807) 2 6=920-44 &'&= 0-61*
S=JTOO =61116-33 c=962-54 c'-c = -0-66#
=N/(1-339600) 2 ^ = 1007-26 d'd=Q-7lx
Av(sepn.) = 920-435 0=3771-00 e'e=-2-59x
10 6 A =27786-57 **=2458-64 '= 2-25*
<5=10 6 A/66 = 421-0087 u=2616-61 t/-v=-2-47*
The links are, of course, in ordinary wave-number units.
It is further suggested that the links may be varied by making them depend on
The Work of Hicks. 57
displacement operations on values of p and s which have already been displaced by
small multiples (x) of 6 or 6 X ; thus, a'=p(xd} p(xd-\-&) ; b'=p(xd A) p(xd),
and so on. The calculated changes of the links are then as shown under a' a, &c.
It will be seen that in a complex spectrum, with seven links and these permissible
variations in each of them, there is room for many accidental coincidences, but the
discussion of probabilities has convinced Hicks that the existence of the links, in
the main, cannot be due to chance.
The following (IV., 366) are among the numerous suggested links occurring in
silver, the differences in wave-numbers of the lines being enclosed in brackets, and
decimal parts of the wave-numbers of the lines being omitted :
(1) 30514 (2460-39) 32974 (2460-84) 35435 (2461-00) 37896 (2457-26) 40353
(2) 17814 (3777-32) 21591 (3779-86) 25371 (3778-01) 29149 (3778-56) 32928
<3) 30959 (3776-44) 34735 (3777-47) 38513 (3773-79) 42286 (2618-00) 44904
Hicks attaches great importance to these long series of the same links, as proving
the reality of variations in the links. Thus, the differences in the second row are
regarded as representing a modified link of about 3778-44 as compared with the
calculated g=3771-00. The normal e link is thus changed to e( 3d) ; that is,
[3771-00+ (3x2-59)] with an outstanding error of 0-33. The link to another line,
P x (l) at 30471 is 2456-59, which is equivalent to w(<5)+0-20.
Proceeding in this way, Hicks has drawn up extensive tables and maps which
are intended to show that a great number of lines may be connected with ordinary
series lines by links and chains, but most of them are too complex for reproduction.
As giving some indication of the nature of the results, however, the short linkage
starting from Ag 1
in order that m=l may give the first line. The diffuse series differ from those of the
alkali metals in exhibiting lines corresponding to m1 (except in the case of Mg)
when IJL is put <1. The numeration of the diffuse series also differs from that
adopted in the alkalis, and is possibly not altogether satisfactory. In the alkalis,
lines of the diffuse series were assigned the same numbers as the lines of the sharp
series to which they were nearest, but to do this in the alkaline earth metals would
require either that the first diffuse triplet should be numbered zero, or that the first
sharp and principal triplet should be given the order-number 2. That the actual
first d triplets have been recorded is indicated by their connection with the
/ series.
In the singlet systems, the component series are arranged in the same way as
in lithium, and the chief difference lies in the appearance of lines corresponding to
m1 in the diffuse series, except in the case of magnesium.
The enhanced lines form systems of pairs, having 4AT for the series constant,
and showing no simple relations to the arc series. These are tabulated under the
headings "ionised Mg " (Mg+), &c. In each case the arc is a sufficient stimulus
to excite some of the enhanced lines quite strongly, but these lines are more especially
developed in the spark (see Plate IV.).
MAGNESIUM.
Mg. At. wt. =24-32; At. No. =12.
The infra-red lines are from observations by Paschen. Of the remaining lines,
most of the wave-lengths are from a paper by Fowler and Reynolds,* but a few are
from measures by Lorenser and by Meggers. In the case of the triplets the limits
adopted for the subordinate series are as calculated by Fowler and Reynolds, cor-
rected to the international scale.
The less refrangible components of the triplets are represented approximately
by the following formulae :
Sl (m) =39760-5 N/(w-|-l-376546 0-062064/w) 2
di(m) =39760-5 N/ (m +0-832086 0-008310/w) 2
* Proc. Roy. Soc., A. 89,]137 (1913).
12
n6
Tables of Series Lines.
Mg TRIPLETS.
CHAP. >:iv.
PRINCIPAL . 1 s m p .
ls=20474-5.
DIFFUSE. Ip md.
1^=39760-5; l 2 = 39801-4 ; l 8 = 39821-3.
X, Int.
v
Av
m
Pw
X, Int.
v
Av m md
5183-67 (10)
72'-70 (9)
19286-0
326-9
40-9
'i
39760-5
39801-4
3838-29 (10) 26045-9
32-31 (10) 086-6
40-7
20-1
(2)
13714-7
67-38 (8)
346-8
39821-3
3229-36 (9)
106-7
3096-91 (9)
32280-9
41-1
15023-3
6654-5
4..1
7^(2)
13820-0
92-97 (8)
322-0
19-8
(3)
7479-5
032-7
50-4
~t i
824-1
91-07 (6)
341-8
032-7
50-4
824-1
f
2851-65 (8)
35057-1
39-6
7657-5
13055-5
53)
7419-0
48-43 (7)
096-7
20-5
(*)
4704-1
46-77 (6)
117-2
6318-55
15822-6
/(4)
4651-9
6319-08
821-3
*
4653-2
2736-53 (7)
36531-8
40-0
33-54 (6)
571-8
19-8
(5)
3229-3
5782-10
17290-0
7j(5)
3184-5
32-06 (5)
591-6
SHARP. Ip ms.
1^=39760-5; l 2 = 39801-4 ; 1^ = 39821-3.
2672-43 (6)
69-56 (5)
68-14 (4)
37408-1
448-3
468-2
40-2
19-9
(6)
2352-9
X, Int.
v Av
m
ms
2632-88 (5)
37970-0
4.1 -ft
5183-67 (10)
72-70 (9)
19 l 8 'o 40-9
19-9
(1)
20474-5
30-04 (4)
28-63 (4)
38011-0
031-3
T-L \J
20-3
(7)
1790-3
67-38 (8)
346-8
2606-64 (4)
38352-1
40-5
3336-69 (8)
32-14 (7)
29961-2
30002-1 ir"'l
(2)
9799-3
03-89 (3)
02-50 (2)
392-6
413-1
20-5
(8)
1408-5
29-9 * 6)
021-9
i
\ /
2588-28 (3)
38624-2
2942-10 (6)
33979-4
85-54 (2)
665-1
(9)
1136-4
38-56 (5)
34020-2 '
(3)
5781-3
84-23 (2)
684-7
36-88 (4) 039-8
2574-93 (3)
38824-4
40-4
* [2781-33
* 78-17
35943-6
984-3
40-7
19-9
(4)
3817-0
72-25 (2)
70-87 (1)
864-8
885-7
20-9
(10)
936-1
* 76-63]
36004-2
2564-91 (2)
38976-0
41-1
2698-13 (5)
95-18 (4)
37051-7
092-3
40-6
19-7
(5)
2709-1
62-21 (1)
60-87 (1)
39017-1
037-5
20-4
(11)
784-2
93-75 (3)
2649-02 (4)
112-0
37738-6
40-6
2557-20 (2)
54-61 (1)
39093-5
133-1
39-6
(12)
667-6
46-17 (3)
44-78 (2)
779-2
799-1
19-9
(6)
2022-1
2551-13 (2)
48-47 (1)
39186-5
227-4
40-9
(13)
574-0
2617-48 (3)
14-65 (2)
38193-3
234-6
41-3
(7)
1567-0
COMBINATIONS.
X
v v calc.
2595-92 (2) 38510-4
93-19 (1) 551-0
40-6
(8)
1250-3
15768-3
15759-1
6340-2 2pi 3^ = 6340-5
6343-9 2p 2 3^ = 6344-6
* Calculated lines, the real lines being
10969-8
9113-5 2p! 4^=9115-9
obscured.
10963-2
9119-1 2^>, 4^ = 9120-0
The Alkaline Earth Metals.
117
Mg TRIPLETS Continued.
FUNDAMENTAL. 2d m/. 2^ = 13714-7.
COMBINATION, lp 2p.
X
V H
*' mf
X
v
v calc.
14877-1
10812-9
6719-9 (
9245-7 (-
J) 6994-8
i) 4469-0
3854-11
3848-09
. 3844-97
3854-53
3848-78
25939-0
979-6
26000-6
25936-2
974-9
Ifa 2p! =25940-5
lp z 2/> 1 =25981-4
1^32^1=26001-3
COMBINATION 1 . lp m/.
X
-
v
v calc.
Ip^pl^lll-t
3051
2833
2729
32766
35288
36633
l^ 1 _4/ = 35291-5
Mg SINGLETS.
PRINCIPAL. 1 5 m P.
15=61672-1.
DIFFUSE. IP mD.
lP=26620-7.
X, Int.
v
m
mP
X. Int.
V
m
mD
2852-11 (10-ff)
2025-82
1828-1
35051-4
49346-6
54702
(1)
(2)
w ;
26620-7
12325-5
6970
8806-75 (8)
5528-42 (8r)
4703-00 (8r)
4351-91 (7y)
4167-39 (6r)
4057-63 (5r)
3986-79 (4y)
3938-43 (3r)
3904-02 (2r)
3878-58 (1)
3859-24 (1)
11351-8
18083-3
21257-1
22972-0
23989-1
24638-0
25075-8
25383-7
25607-4
25775-3
25904-5
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
15268-9
8537-4
5363-6
3648-7
2631-6
1982-7
1544-9
1237-0
1012-3
835-4
716-2
SHARP. IP mS.
!P=26620-7.
X, Int.
' v
m
mS
2852-11 (10-ff)
11828-8
5711-09 (4)
4730-16 (2)
4354-36 (1)
35051-4
8451-7
17504-9
21135-0
22959-1
(1)5-
(2)
(3)
(4)
(5) ,
61672-1
18169-0
9115-8
5485-7
3661-6
Unclassified Lines of Mg.
X, Int.
v
COMBINATION. 25 mP.
25 = 18169-0-
23991-3 (15)
23977-1 (8)
23963-6 (5)
17073-1
11054-2? (15)
7800-4 (1)
7779-9 (1)
6332-26 (2)
6021-70 (1)
2782-98 (6)
81-52 (6)
79-84 (6)
78-29 (6)
76-76 (6)
4167-0
4169-5
4171-9
5855-6
9043-9
12816-4
12850-1
15787-8
16602-0
35922-1
941-0
962-7
982-7
36002-5-
X
v
v calc.
11828-8
*17108-1
8928-97
- 8451-7
5843-6
11196-4
2S !P=-8451-7
25 2P= 5843-5
25 3P= 11199
COMBINATION. IP mP.
X
v
v calc.
4511-2
4250-8
4106-6
4018-1
22160-9
23518-4
24344-2
24880-4
lp_4P|=22162-6
IP 5P- =23528-3
IP 6Pt=24351-5
IP 7Pf=24885-2
* J chanson includes this in the Rydberg series
as IP+ID, giving ID =32464-3.
f These are terms calculated by Lorenser.
The lines were observed by Fowler, and are all
faint and diffuse.
I Used in calculation of 15. This line occurs
in the flame spectrum and is strong in the electric
furnace at low temperatures.
These closely represent a (s) triplet, but .other
lines are probably involved.
Mg INTER-COMBINATIONS.
X V
v calc.
J4571-15 (4)
12083-2
9257-9
3043-75
2768-47
2765-34
21870-7
8273-7
10798-6
32844-7
36110-3
36151-2
15 l a =21870-7
2D3f = 8274-1
2D 4/ =10799-9
Ip 3 3P=32850-8
\p-i- 5Z>=36111-8
lp 2 5D =36152-7
ii 8 Tables of Series Lines. CHAP. xiv.
The residuals C(Av) are as follows :
m 1 2 3 456789 10 11 12 1.3
s 0-0 0-0 0-5 0-8 0-8 0'6 -0-8 0-5
d 0-0 0-0 2-0 2-3-1-7 2-0 2-0-1-5 -1-2 1-0 0-80-7
In the formula for s, /m has been put >1 so that m=\ gives a positive value, in
accordance with the occurrence of the p series on the less refrangible side of s. If jn
were < 1 the formula would suggest a p series in the ultra-violet, but this has not
been observed.
Direct calculation of the limit for the diffuse series of singlets gave 26619-2, and
thence, by adding the wave-number of the first principal line, 61670-6 for the limit
of the principal series. The limits adopted, however, have been derived from the
combination IS l 2 =21870-7, which gives 61672-1 for IS, and thence 26620-7
for IP.
The combinations have been taken from the works of Dunz and Lorenser.
Attention has already been drawn (p. 37) to the difficulty of representing the
singlet D series by a formula.
IONISED MAGNESIUM (Mg+).
The enhanced lines of magnesium have been the subject of an extended in-
vestigation by Fowler.* The spectrum is remarkable for the fact that some of
the enhanced lines appear conspicuously in the ordinary arc in air while others
associated with them in their series relations only appear in the spark, or in the
equivalent conditions in an arc in vacuo (see also p. 66). The lines which appear
in the arc in air are U 2936, 2928, 2802, 2795, 2798, 2790, constituting three pairs
with a separation in wave-number of 91-5. Two pairs of like separation have been
observed in the spark by Lyman at 1753-6, 1750-9, and 1737-8, 1735-0 ; it is not
certain that these appear under ordinary arc conditions.
In the spark in air the majority of the arc lines are present, but the first three
pairs above mentioned are strengthened, and there is an intense broad line at
A4481 (see PI. IV.), together with feeble indications of other extremely nebulous lines,
including one about A3105. As observed in the arc in vacuo, ^4481 becomes well-
defined, and many other lines appear which are probably so diffuse in the spark
as to escape detection. Among the additional lines are a number of doublets having
a separation of 30-5, and two series of single lines. Under the conditions of the
vacuum arc, the line 4481 is revealed as the leader of a strong series which extends
far into the ultra-violet ; with high resolution, the first three members have been
shown to be very close doublets, having the stronger components on the more re-
frangible sides, and a constant separation Ai> of 0-99. The other members of the
series are too close for resolution with the grating. This series is of the fundamental
( 2 .
Formulae for the individual series were given by Fowler, and served to justify
the arrangement of the lines adopted in the accompanying table. All the observed
enhanced lines are included in the scheme. In drawing up the table, the limits
49777-0 and 49776-0 have been adopted for the " 4481 " series, and all the other
limits have been derived from these with the aid of observed lines, whether occurring
in the main series or in combinations. Thus, starting with 2<5=49777'0, we have
In 1= 2d+d(2) =49777-0+35729-44=85506-44 ; and 10=1^+^(1) =85506-44 +
35760-97=121267-41.
* These have sometimes been designated the " FP " (Fowler and Payn) doublets- of
magnesium.
f Used in calculation of constants.
I2O
Tables of Series Lines.
Mg + DOUBLETS.
CHAP. XIV.
PRINCIPAL, lo mis.
10 = 121267-41.
COMBINATION, TC TYPE. 2o WITT.
2o=51462-2.
X, Int.
V
Av
m
W7T 1)2
X, Int.
v
Av
m
m7t ll2
2795-523 (50)
2802-698 (50)
1239-9*
1240-4*
1026-0*
1026-1*
35760-97
35669-42
80651-3
80620-8
97469-0
97454-9
i
91-55
30-5
14-1
(1)
(2)
(3)
85506-44
597-99
40616-1
646-6
23798-4
812-5
2936-496 (35)
2928-625 (35)
9217-4*
43-4*
3613-80 (4)
15-64 (3)
2790-33*
90-92*
34044-25
34135-74
10846-1
10815-6
27663-82
27649-75
35827-5
35819-9
91-49
30:5
14-07
7-6
(1)
(2)
(3)
(4)
85506-45
597-94
40616-1
646-6
23798-4
812-5
15636-7
644-3
SHARP. ITT mo.
17^=85506-44; lTT 2 = 85597'99.
X, Int.
v Av
m
mo.
COMBINATION, 8 TYPE. 27i mS.
27^=40616-1 ; 27u a = 40646-6.
2795-523 (50)
35760-97
35669-42
34044-25
34135-74
57027
57115
91-55
91-49
88:
(1)
(2)
(3)
121267-4
51462-2
28481-2
2802-698 (50)
2936-496 (35)
28-625 (35)
1753-6
50-9
X, Int.
v
Av
m
r8
*
*
7896-37
77-13
4390-585 (10)
84-643 (8)
3538-86 (6)
35-04 (5)
3168-98 (3)
65-94 (2)
2967-87 (1)
65-19 (0)
9130-4
9160-9
12660-6
12691-5
22769-64
22800-50
28249-6
28280-1
31546-8
31577-1
33684-4
33714-9
30-5
30-9
30-86
30-5
30-3
30-5
(2)
(3)
(4)
(5)
(6)
(?)
(8)
49777-0
27955-3
17846-3
12366-5
9069-4
6931-7
5471-7*
DIFFUSE. ITT m8.
1^ = 85506-44; l7t 2 = 85597-99.
X, Int.
v Av
m
mS, 8'
2797-989 (40)
t
90-768 (40)
1737-8
35-0
35729-44
35730-43
35821-90
57546
57639
0-99
91-46
93:
(2)
(3)
49777-0
776-0
27960 :
COMBINATION, o TYPE. 27r wo.
27^=40616-1 ; 27r 2 =40646-6.
X, Int.
v Av m ; ma
28 "1
COMBINATION,
4481 -327 \
81-129/ (1
3104-805 \
04-713/ (30)
2660-821 \
60-755 / (1U
+2449-573 (6)
J2329-58 (5)
J2253-87 (4)
{2202-68 (3)
2166-28 (2)
22308-58
22309-57
32198-83
32199-79
37571-22
37572-15
40811-08
42912-93
44354-35
45385-03
46147-60
0-99
0-96
0-93
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
27467-4
17577-2
12204-8
8965-6
6863-8
5422-3
4391-7
3629-1
3049-0*
* Calculated lines or terms,
t Probable satellite.
J Not resolved ; 49776-7 adopted for limit.
The Alkaline Earth Metals.
121
Mg+ DOUBLETS Continued.
COMBINATION. 38 w 9.
38=27955-3.
COMBINATION. 89 w 9 (?)
39=27467-4.
X, Int.
v obs.
v calc.
m
X, Int. v obs;
v calc.
m
6346-67 (5)
15752-0
15750-5
(5)
6545-80 (5)
15272-8
15262-6
(5)
5264-14 (5)
18991-2
18989-7
(6)
5401-05 (5)
18509-8
f 8501 -8
(6)
4739-59 (5)
21093-0
21091-5
(7)
4851-10 (5)
20608-2
20603-6
(?)
4436-48 (5)
22534-1
22533-0
(8)
4534-26 (4)
22048-1
22045-1
(8)
4242-47 (4)
23564-6
23563-6
(9)
4331-93 (3)
23078-0
23075-7
(9)
4109-54 (3)
24326-8
24326-2
(10)
4193-44 (2)
23840-1
23838-3
(10)
4013-80 (2)
24907-0
24906-3
(11)
4093-90 (1)
24419-7
24418-4
(11)
COMBINATION. 28 3TT 1)2 .
28=49776-7.
X, Int.
v obs.
v calc.
Av
3848-24 (7)
25964-0
25964-2
3850-40 (6)
25978-6
25978-3
14-6
CALCIUM.
Ca. At. wt. =40-07; At. No. =20.
The arc spectrum of calcium includes a system of triplets and a system of
singlets. Lines of a doublet system also occur in the spectrum of the arc, but these
ibelong to ionised calcium (Ca +) and will be considered separately.
Data relating to the arc lines have been collected and much extended by Saunders,*
.and the series are tabulated as given by him. Many of the adopted measures were
'made by Crew and McCauley,f who used the vacuum arc, and others were made by
_A. S. King in the course of his work with the electric furnace. Preliminary observa-
tions of the infra-red region were provided by H. M. Randall. A valuable set of
wave-lengths has also been given by Holtz.J
The p series of triplets lies in the infra-red, and has only been partially recorded ;
'but the s, d, and / series are well marked in the visible and ultra-violet regions. In
the d series, the separations of the satellites from the chief lines show marked irregu-
larities, the usual shrinkage continuing only to the fourth member, after which the
separations increase ; the satellites, however, show constant separations from the
third line characteristic of the triplet, and it is therefore the chief components which
are abnormally displaced. Reference has already been made to the difficulty of
representing the d series by formulae (p. 42). The separations in the / triplets
show the usual correspondence with those of the satellites of the first d triplet.
There are indications that the triplets of the / series have satellites, as in the case of
ibarium, but the lines are too close for measurement in any but the first.
The P series of singlets begins with the well-known flame line 4226A, and all
the lines are easily reversed, except the second, which is abnormally faint. This
does not reverse, as is also the case with the corresponding lines in Sr and Ba, but
* Astrophys. Jour., 52, 265 (1920).
t Astrophys* Jour., 39, 29 (1914).
j Zeit. Wiss. Phot., 12, 201 (1913).
122
Tables of Series Lines.
Ca TRIPLETS.
CHAP. XIV.
PRINCIPAL . 1 5 mp .
ls = 17765-l.
FUNDAMENTAL. 1 d mf.
1^=28933-5; l^'=28955-2 ; ld"=28968-8.
X, Int. v
Av
m
#!...,
X, Int.
V
Av
m
mf
6162-18(9) 16223-6
105-9
33988-7
4585-92 (2)
21799-7
(3)
7133-9
- 6122-22(8) 329-5
52-2
(1)
34094-6
85-87 (6)
799-5
- 6102-72(8) 381-7
146-9
21-5
4581-41 (5)
21821-2
13-6
19856-3 5034-8
19-9
*12730-3
4578-57 (4)
21834-8
19935-2 5014-9
(2)
750-2
4098-55 (4)
24392-1
21-5
(4)
4541-5
f!0987-3
7-8
6777-8
94-94 (3)
413-6
13-7
f 979-5
4-0
(3)
85-6
92-65 (2)
427-3
t 975-5
89-6
3875-81 (4)
25793-9
21-8
(5)
3139-5
72-55 (3)
815-7
13-5
fl 3422-4
(4)
4342-7
70-51 (2)
829-2
SHARP. Ip ms.
1^=33988-7; l/> 2 =34094-6 ; l/> 8 = 34146-9.
3753-37 (1)
50-35 (1)
48-37 1)
26635-3
656-8
670-9
21-5
14-1
(6)
2298-1
X, Int.
V
Av m
ms
\ /
3678-24 (2)
27179-4
21-6
(7)
1754-1
6162-18 (9)
22-22 (8)
16223-6
329-5
105-9
52-2
(1)
17765-1
75-31 (2)
73-45 (1)
201-0
214-8
13-8
02-72 (8)
381-7
3628-60
27551-2
22-1
(8)
1382-3
3973-72 (4)
57-05 (5)
25158-4
264-3
105-9
52-2
(2)
8830-3
25-69
24-11
573-3
585-3
12-0
48-90 (3)
316-5
3594-08
27815-7
21-8
(9)
1117-7
3487-61 (5)
74-77 (3)
28664-9
770-9
106-0
52-1
(3)
5323-8
91-26
89-49
837-5
851-3
13-8
.
68-48 (2)
823-0
3568-91
28009-6
24-3
(10)
923
3286-06 (3)
30423-1
105-8
66-12
033-9
13-9
74-66 (2)
528-9
52-1
(4)
3565-6
64-35
047-8
69-09 (1)
581-0
3550-03
28158-5
23-5
(ID
774
3180-52 (2)
69-85 (1)
31432-5
538-3
105-8
52-1
(5)
2556-2
47-38
45-58
182-0
196-3
14-3
^
64-62 (1)
590-4
Oil 7. *(* /I \
3535-55
28273-8
(12)
660
oil /*OO (1)
ojJUbo'o
105-9
07-39 (1)
02-36 (0)
172-2
224-4
52-2
(6)
1922-4
* Terms calculated from combination Id 2p
are 12729-4, 12749-5, 12756-4.
3076-99
67-01
62-05
32490-1
595-6
648-5
105-7
52-7
(7)
1498-6
t Calculated, not observed.
J Probably belongs to d series.
See also d series.
3049-01
32788-2
105-8
39-21
894-0
1200-3
J34-52
944-8
3028-97
33005-2
19-37
110-1
10 ^ 9 (9)
982-5
3014-01
33168-9
(10)
819-8
The Alkaline Earth Metals.
Ca TRIPLETS Continued.
123
DIFFUSE, ipmd.
1^=33988-7; l 2 =34094-6 ; 1^=34146-9.
X, Int.
V
Av
m
md", d', d
X, Int.
V
Av
m
md", d', d
19917-3
5019-6
11-3
(1)
28969-1
(7)
1848-9
864-3
771-1
32-9
55-1
J. tl O
22-2
955-2
933-5
3109-51
3108-58
32150-3
159-9
9-6
38-7
28-8
105-6
105-4
19506-8
5125-2
452-6
39-5
14-3
3099-34
32255-7
52-1
19310-3
5177-3
3095-29
32298-0
4456-61(3)
22432-4
0.7
(2)
11556-4
(8)
1551-2
55-88(5)
436-1
> i
K.K
552-6
3081-55
32441-9
7-7
47-0
54-77(9)
441-7
o u
547-0
80-82
449-6
39-1
105-9
105-4
4435-67(8)
22538-3
3-7
3071-97
32543-2
4-1
34-95(9)
542-0
71-58
547-3
52*1
52-6
4425-43(9)
22590-4
3067-01
32595-8
3644-99(0)
27427-3
1 .7
(3)
6561-4
(9)
1272-7
44-76(3)
429-0
It
2-8
59-7
3055-55
32718-2
2-3
70-7
44-40(7)
431-8
56-9
55-32
720-5
68-2
106-0
105-1
3630-97(2)
27533-3
1-6
30-75(6)
534-9
3045-75
32823-3
52-0
3624-11(6)
27585-3
3041-05
32874-0
3362-28(0)
29733-3
1-4
(4)
4255-5
(10)
1045-4
62-13(2)
734-7
i.a
54-0
61-92(6)
736-5
A O
52-2
3034-52
32944-8
105-8
3350-36(2)
29839-1
1 -1
50-20(5)
840-6
i CF
3024-93
33049-2
52-2
3344-51(5)
29891-3
3020-15
33101-5
(5)
3002-4
3018-55
33119-1
(11)
869-6
3226-13(1)
30988-1
2-4
00-6
25-88(5)
990-5
%
2998-2
105-8
3006-22
33254-9
(12)
733-8
3215-33(1)
31092-2
1-7
15-15(3)
093-9
52-3
2996-67
33360-8
(13)
627-9
3209-93 (3)
31144-5
2988-98
33446-7
(14)
541-0
(6)
2268-2
3151-28(1)
50-75(4)
31724-1
729-4
5-3
64-5
59-3
2982-89
33515-2
(15)
473-5
106-0
3141-16(0)
31826-3
3-8
40-78(2)
830-1
52-4
.3136-00(2)
31878-7
124
Tables of Series Lines.
Ca TRIPLETS Continued.
CHAP. XIV.
COMBINATIONS. Id mp.
COMBINATION. 2d mf. ',
' X, Int.
V
v calc.
X
V
v calc.
6169-58(3)
16204-1
\d-2p! = 16204-1
22655-9
4412-9
2d 3/ = 4413-1
6161-31(2)
225-9
ld'2p j = 225-8
624-6
19-0
2d' 3/ = 4418-7
6156-08
239-7
Id"2p 1 = 239-4 610-0
21-8
2d"3f = 4422-5 ,
6169-03(2)
205-6
Id' 2p 2 = 205-7
i
i
6163-75(2)
ftl flA.J_A/^M
219-5
91 9. A
Id" 2p 2 = 219-3
1/7" 9>i 91 9. A.
14278
7002
2d 4/ = 7005-5
4512-28(1)
22155-7
ld 3p!= 22155-7
07-85(0)
177-4
Id' 3pi= 177-4
05-00
191-4
Id"3p 1 = 191-0
COMBINATION. 2pwd. , ,
09-45(0)
06-62
07-42
169-6
183-5
179-5
Id' 3p 2 = 169-6
Id" 3^2= 183-2
1^" 3 3 = 179-2
X i Possibly part of incom-
16200-0 pletely observed group
162-2 J" of six lines of combina-
4065-44
24590-8
144-8 J tion2p 3d.
nd 4/>! = 24590-8
4062-49
608-7
?ld' 4j = 24612-5
Ca SINGLETS.
PRINCIPAL. IS mP*
15=49304-8.
DIFFUSE. IPmD.
!P=25652-4.
X. Int.
v
m 1 mP
X, Int.
v
m
mD
4226-73(10^)
f2721-65
2398-58(2)
2275-49(1)
2200-78(1)
2150-78(1)
2118-68
2097-49
2082-73
2073-04
2064-77
23652-4
36731-8
41678-8
43933-3
45425-0
46480-5
47184-7
47661-3
47998-9
48223-2
48416-3
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
25652-4
12573-1
7625-9
5371-4
3879-6
2824-6
2120-3
1638-2
1305-9
1071-6
888-5
7326-10(8)
5188-85(3)
4685-26(2)
4412-30
(-1802-9)
13646-1
19266-9
21337-7
22657-7
(1)
(2)
(3)
(4)
(5)
27455-3
12006-3 ,
6385-5
4314-7
2994-7
FUNDAMENTAL. ID mF. J
ID =27455-3.
X, Int.
v
m
mF
4878-13(5)
4355-10(5)
4108-55(1)
3972-58(1)
3889-14(1)
3833-96
3795-62
3767-42
20494-0
22955-3
24332-7
25165-6
25705-5
26075-5
26339-0
26536-0
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
6961-3
4500-0
3122-6
2289-7
1749-8
1379-8
1116-3
919-3
SHARP. IP mS.
!P=25652-4.
X, Int.
v
m
mS
4226-73(10^)
10345-0
5512-98(4)
4847-29(2)
4496-16
4312-31
4203-22
4132-64
4084-5
23652-4
9664-2
18134-0
20624-4
22235-1
23183-0
23784-7
24190-9
24476
rof
(2)f
(3).V
(4)
(5)
(6)
(7)
(8)
(9)
49304-8
15988-2
7518-4}
5028-0
3417-3
2469-4
1867-7
1461-5
1176
COMBINATION. ID mP.
X, Int.
v
v calc.
6717-69(5)
5041-61(3)
4526-94(4)
4240-46(2)
4058-91(0)
3946-05(0)
3871-54
(1803)
14882-0
19829-5
22083-9
23575-8
24630-4
25334-8
25822-3
]
]
'
WIP= 1802-9
ID 2P 14882-2
* Formerly known as SL,1 (81,=" single
line"). A strong series.
t Abnormally faint.
' j Formerly known as SL3.
Formerly known as SL/2. Faint and diffuse
in air, but of medium strength in vacuum arc.
LD 3P = 19829-4
ID 4P= 22083-9
W5P =23575-7
LD 6P =24630-7
LD7P =25335-0
LD 8P=25817-1
The Alkaline Earth Metals.
125
Ca SINGLETS Continued.
COMBINATION. IS mD. *
Unclassified Triplets and Pairs of Ca.
X
V
v calc.
X, Int. v
Av
4575-43
2680-36
2329-33
2221-91
21849-9
37297-6
42918-0
44992-6
IS ID =21849-5
IS 2D 37298-5
IS 3Z> 42919-3
IS 4D 44990-1
4318-648(9)
4298-989(8)
89-363(9)
5601-283(5)
5594-464(8)
90-109(5)
5270-272(5)
64-237(3)
60-375(1)
7202-161(2)
7148-123(3)
4302-525(9)
4283-008(9)
3006-864(5)
2997-309(4)
6499-648(4)
93-789(8)
6471-659(4)
62-576(9)
6455-606(3)
49-811(7)
5602-829(5)
5598-484(8)
5588-746(9)
81-973(6)
5265-559(5)
61-701(3)
23148-91
254-76
306-96
17848-11
869-87
883-79
18969-08
990-84
19004-77
13880-89
985-83
23235-66
341-53
33247-58
353-57
15381-20
395-08
15447-72
469-44
15486-15
500-05
17843-18
857-03
17888-15
909-85
18986-07
999-99
105-85
52-20
21-76
13-92
21-76
13-93
104-94
105-87
105-99
13-88
21-72
13-90
13-85
21-70
13-92
COMBINATION. IS mS.
X
V
v calc.
2392-22
2257-40
2177-8
41789-8
44285-5
45883
IS 35=41786-4
1545=44276-8
1555=45887-5
OTHER COMBINATIONS.
X, Int.
V
v calc.
7645-25(1)
4929-25
13076-9
20281-4
3412
8359
19935-2
IP 2P = 13079-3
IP 4P=20281-0
25 2P= 3415
25 3P= 8362
ID 35 = 19936-9
Ca INTER-COMBINATIONS.
X
V
v calc.
6572-78(3)
2734-82
3761-70
15210-3
36554-9
26576-2
lSl^ 2 = 15210-2
IS 2/> 2 = 36554-6
Ip 2 35=26576-2
Possibly also ID 2p lt lD2p 3 , !D3p 3
Other Unclassified Lines of Ca.
X, Int. v Av
* Faint and diffuse except in vacuum sources ;
last two wave-lengths difficult to determine.
6439-086(9) 15525-87
5857-476(8) 17067-48
5349-470(5) 18688-25
5262-238(3) 18998-05
4307-738(7) 23207-53
4298-989(8) 23258-17
3000-865(4) 33311-50
126 Tables of Series Lines. CHAP. xiv.
the line must belong to the series, because its " term " enters into combinations
which appear in the spectrum (see also p. 40). The terms quoted for the P series
are averages from P itself and the combination IDmP. Saunders remarks that the
courses of the S and D series would have been difficult to trace but for the existence
of parallel combination series. In the D series it should be noted that the suggested
first line has a negative frequency ; it seems necessary to suppose its existence in
order to give an appropriate connection with the F lines. Support for the selected
F series is given by the close approach of the terms to those of hydrogen (i.e., ^
is nearly unity), which is a general feature of such series.
The limits chosen by Saunders have been adjusted to accord with the inter-
relations indicated by the various series and combinations, and are believed to have
been found with considerable precision.
In addition to combinations in each of the two groups of series, there are
lines which result from combinations of terms taken from the singlet and triplet
systems. As Saunders remarks, this points to a close relation between the two
systems.
There are several interesting triplets and pairs which have not been found to
belong to the regular system, although their separations are identical with those
observed in the series. These and the brighter unclassified single lines have been
included in the tables, as further investigation of possible series or combinations is
very desirable.
The intensities shown in the table are those given by Crew and McCauley, as
photographed in the vacuum arc. The intensities of the additional lines observed
by Saunders have not been stated, but it may be assumed that all the lines for which
no intensities are quoted are of low intensity.
IONISED CALCIUM (Ca+).
The enhanced lines of calcium form a system of pairs, of which but few occur
outside the Schumann region. The well-known H and K lines form the principal
pair. Crew and McCauley's wave-lengths have been adopted in the range which
they cover, and the remainder are as given by Lyman, except 6(1), in which case
the wave-lengths are the means of the values given by Meggers and Meissner. The
separation of the main pairs is 222-85, and that of the narrower fundamental pairs,
as indicated by the satellite separation in 6 (1) is 60-85.
The limit 70325-29 was calculated by a Hicks formula from the less refrangible
components of the first three CT pairs, giving the formula :
Ol (m) =70325-29 -4JV/(w+l-205543-0-064899/w) 2 .
The C ( Av) equal +0-01,* +0-04,* 0-00,* 18, 24. As shown by Hicksf
a more accurate representation of the series is obtained by putting / u>l, and the,re
is the additional advantage that m=l then gives the first pair. The first pair of
6 occurs with negative sign, and this also is given by m=\, when ju, in that series
is taken to be greater than unity. No combination lines have yet been
recognised.
* Used in calculation of constants.
f Proc. Roy. Soc., 91, 452 (1915).
The Alkaline Earth Metals.
Ca + DOUBLETS.
127
PRINCIPAL.
lo mn.
DIFFUSE. ITC m8.
lo=95739-70.
17^ = 70325-29 ; lTU 8 = 70548- 14.
X, Int. v
Av
m
nci,.
X, Int.
V
Av
m
w8', 8
3933-664(10fl) 25414-41
3968-465(10^?) 191-56
222-85
(1)
70325-29
548-14
8498-00 (8)
8542-15 (10)
11764-25
703-44
60-85
222-89
(1)
82089-52
028-73
QfJfJO.l 1 /Q\
KA] .Ofi
SHARP. ITT mo.
17^ = 70325-29; I7r 2 = 70548-14.
3181-283 (6)
79-340(10)
58-877(10)
31424-80
443-99
647-69
19-19
222-89
(2)
38900-47
881-30
X, Int.
V
Av
m
wo
3933-664(10.ff)
25414-41
191-56
222-8
(1)
95739-70
8-61
222-89
(3)
23017-56
008-95
3968-465(10^)
2112-763 (2)
47316-34
03-239 (2)
530-58
3736-903 (9)
06-022 (9)
26752-55
975-46
222-91
(2)
43572-71
1815-0
55096
(4)
15232
07-8
316
2208-606 (3)
2197-791 (3)
45263-26
485-97
222-71
(3)
25062-10
1680-5
59506
(5)
10815
74-1
733
1851-3
43-8
54016
236
220
(4)
16310
FUNDAMENTAL. 18 m<$.
1698-9
58861
18 = 82029 ; l8'=82090.
92-4
59087
226
* I
12-1
SHARP. | \p ms.
1^ = 31038-0; l^ 2 =31432-2; l 3 =31619-0.
-' /, Int. v
Av m
ms
3867-2
25851-2
(?)
1766-0
7070-10)7) j 14140-2
,
6878-35 (7) 534-4
186-8
16897-8
DIFFUSE, ip md.
6791-05(6) j 721-2
1/> 1 =31038-0 ; l/> 2 = 31432-2 ; 1/> 3 = 31619-0.
4438-04 (4) i 22526-2
004.0
X, Int.
v Av
m
md", d', d
4361-71 (4) ! 920-4
(9\
1 /* 1 '
8511-8
4326-44(3) j 23107-2
186-8
(3260-4)
"f\ (\
(1)
27777-6
30109-7
3320-3
:>9-y
717-7
3865-46(4) 25862-8
3Q4.-5
29225-0
3420-8
LOO-5
617-2
07-38(4) ! 26257-3
Ov*T 2
X, Int.
V
Av
m
mS', 8
5350-46 (\OR) 18684-8
7792-7
(1)
41471-5
3529-43 (8R)
28325-2
82-0
(2)
13146-2
3775-72 (10J2)
26477-5
49264-2
19-24 (10.R)
407-2
064-3
11513-22
13013-8
8683-3
7682-1
1001-2
(2)
14103-4
15104-6
2767-87 (10 R)
36118-2
7793-0
6549-77 (8)
15263-5
372-7
(3)
7523-2
2921-52 (QR)
34218-8
37-5
(3)
7252-8
6713-69 (6)
14890-8
7895-9
18-32 (10 R)
256-3
15-2
5527-90 (4)
83-98 (2)
18085-0*
17903-4
181-6
(4)
4701-7
4883-3
2379-58 (8R)
42011-3
7792-5
5109-47 (2)
19566-1
IAJ..O
(5)
3220-6
2710-67 (45)
36880-4
19-6
(4)
4591-6
36-84 (I)
461-8
1U4 o
3324-9
09-23 (SR)
900-0
71-5
7791-6
4891-11 (2)
4906-3 (2)
20439-6
376-3
63-3
(6)
2347-1
2410-4
2237-84 (6R)
44672-0
4760-6 (1)
20999-9
QA.fi
(7)
1786-8
2609-77 (4R)
38306-1
11-5
(5)
3165-8
68-5 (1)
965-1
O*
1821-6
08-99 (6fl)
317-6
53-9
7791-9
4678-1 (3)
4617-2 (2)
21370-2
21652-1
(8)
(9)
1416-5
1134-6
2168-61 (4:R)
46098-0
4574-6 (1)
4547-9 (0)
21853-7
21982-0
(10)
(U)
933-0
804-7
f2552-98 (ZR)
52-53 (QR)
39158-1
165-1
7-0
(6)
2314-7
06-4
SHARP. ITC ma.
7790-0
17^=41471-5 ; l7t 2 =49264-2.
2129-33 (IR)
46948-1
X, Int.
V
Av
m
ma
2517-41 (4#)
39711-4
(7)
mo
1760-1
J2105-0
47491
5350-46 (107?)
3775-72 (10.R)
18684-8
26477-5
7792-7
(1)
22786-7
2493-91 (2R)
40085-6
(8)
1385-9
2088-7
47861
\ /
3229-75 (10.R)
2580-14 (8R)
30953-2
38746-0
7792-8
(2)
10518-3
2477-49 (Ifl)
40351-2
(9)
1120-3
2077-2
48126
\ v l
2826-16 (8R)
2315-93 (QR)
35373-3
43166-0
7792-7
(3)
6098-2
2465-46 (IR)
2069-1
40548-0
48315
(10)
923-5
2665-57 (2w)
2207-06 (4#)
37504-3
45295-0
7790-7
(4)
3968-2
2456-45 (IR)
2062-2
40696-9
48476
(U)
774-6
2585-59 (4fl)
2152-01 (IR)
38664-4
46453-5
7789-1
(5)
2808-9
2449-49 (IR)
2057-2
40812-5
48594
(12)
659-0
2538-18 (2R)
12119-1
39386-5
47174-6
(6)
2085-0
2443-92 (IR)
2053-8
40905-8
48675
(13)
565-7
2507-94 (IR)
2098-4
39861-3
47640-1
(7)
1610-2
2439-50 (1.R)
40979-6
(14)
491-9
2487-48 (IR)
2083-1
40189-2
47989-9
(8)
1282-3
FUNDAMENTAL. 28 mcp.
28=13064-3; 28' = 13146-2.
2472-57 (IR)
OHTO.Q
40431-5
<4Q9JiV.9
(9)
1040-0
X
V
Av
m
my
-U /Z o
*2461-93 (IR)
*&6'.\J-*
40606-2
(10)
865-3
16340-3
16123-0
6118-2
6200-7
82-5
(3)
6945-8
*2453-79 (1.R)
40741-0
(U)
730-5
*2447-51 (IR)
*2442-16 (IR)
40845-5
40935-0
(12)
(13)
626-0
536-5
11594-5
11482-2
8622-4
8706-8
84-4
(*)
4440-7
* More refrangible components not recorded.
t Satellites not observed beyond this pair.
(5)
[3077-0]
J This and smaller wave-lengths may be much
in error.
9170-7
10901-3
(6)
2244-9?
M
l62
Tables of Series Lines.
Tl COMBINATIONS.
CHAP. XVII.
COMBINATIONS lo mS.
lo=22786-7.
COMBINATIONS 2^ mS.
X
v
v calc.
X, Int.
v
v calc.
51057-9
14592-6
12736-4
10496-4
9512-4
9136-1
8376-1
14515-5
10492-5
1958-0
6851-0
7849-4
9524-5
10509-7
10942-6
11935-5
6887-3
| 9528-1
10292-3
6420-45 (1)
5488-79 (2)
5093-28 (1)
9713-4
15570-9
18213-9
19628-2
lo 28= 9722-4
1038 = 15571-5
1048 = 18215-2
lo_58 = 19632-8
27^38' =
27t 2 38' =
2^48' =
2rr 2 48' =
1958-4
= 6850-6
= 7851-8
= 9511-8
= 10513-0
COMBINATIONS nm mo.
27T 2 58' =
= 10937-6
= 11938-8
X
v
v calc.
27TJ 38 -
= 6888-2
27889-6
21803-0
12491-8
7-023 \L
5-559[z
3584-6
4585-3
8003-7
1423-5
1798-6
2:r 1 2o = 3585-1
27T 2 2o = 4586-3
27T! 3o-8005-2
27T! 48 =
= 9531-9
COMBINATIONS 38 my.
37^30 = 1425-0
37T 2 3o 1797-7
X
v
v calc.
3-505 p
3-568[z
2781
2803
38 49=2774-5
38' 49=2812-1
COMBINATIONS ITT ;WTT.
X, Int.
v
v calc.
Unclassified Lines of Tl.
*3652-95 (In)
2945-04 (4r)
f2719-10 (1)
2977-93 (lr)
f2843-27 (1)
2416-70 (lr)
27367-3
33945-5
36766-1
33570-6
35160-5
41366-1
l7T 1 27r 1 =27368-1
iTij 37T! =33948-3
l7T 1 4TC 1 = 36769-8
iTTj 3ru 2 = 33575-6
lTT 2 27^ =35160-8
l7T 2 37t 2 =41368-3
X
v
X
v
7-117(z
3-92865 {Jt
3-92155
2-70276?
2-70237
2-13979
2-04858
1-45978
1-16907
3230-6 (1)
2210-73 (2R)
1404-7
2544-7
2549-3
3698-9
3699-4
4672-1 !
4880-1 ;
6848-5
8551-4
30945-1
45219-8,
J2671-10 (2w)
2669-95 (1)
2577-67 (1)
2532-71 (In)
2530-80 (In)
2512-59 (1)
2434-05 (1)
2417-01 (2)
37426-7
37442-8
38783-1
39471-6
39501-3
39787-5
41071-3
41360-8
COMBINATIONS ITT my.
X, Int.
v
v calc.
2895-41 (4)
2362-08 (2v)
2700-2 (2n)
34527-3
42322-7
37023-3
iTTi 39 = 34525-7
I7t 2 39 = 42318-4
iTtj 49 = 37030-8
COMBINATIONS 2o mn.
2o = 10518-3.
* Not given by Dunz.
t Given by Huppers.
j This and succeeding lines are given by
Huppers, not by Kayser and Runge. The
first, second, fifth, and seventh are given
in spark by Eder and Valenta.
X
v
v calc.
33393-2
38131-0
2993-8
2621-8
2o STTJ =2995-1
2o37T 2 =2622-4
ordinate series were calculated by Dunz from the sharp series. It will be seen that
nearly all the lines are included in the recognised series.
The enhanced lines do not appear to have been investigated for series relation-
ships. The vacuum-arc and spark spectra in the region A1908 to A1477 have been
observed by McLennan.*
The sharp series is not well represented by most of the ordinary formulae. For
wave-numbers on the Rowland system, however, Johanson has obtained a fair
agreement by his formula (see p. 36) with the constants A =41469-33 ; ^=2-261774 ;.
a=l-672006: b = 1-937964.
* Proc. Roy. Soc., A. 98, 108 (1920).
CHAPTER XVIII.
ELEMENTS OF GROUPS IV. AND V.
ELEMENTS OF GROUP IV.
There are no published records of the discovery of typical series in the spectra
of any of the elements of this group. Constant difference groups, however, were
found by Kayser and Runge in the spectra of lead and tin which have been verified
with remarkable accuracy in Kayser's laboratory by Klein* and Arnolds f
respectively. In the arc spectrum of lead there are 10 lines differing in wave-number
by 10807-4:3 from 10 other lines, and these, again, by 2831-2 from other more
refrangible lines. Similarly, for the arc lines of tin there are 10 sets of three lines
for which the separations are 5185-43 and 1735-84.
In the spectrum of tin LohuizenJ has suggested two groups of parallel series,
which he has called " translation series." Thus, he gives six series with limits
45307-40, 50494-43, 50926-14, 52330-66, 53507-27, 53924-00, for all of which the
terms " mx " are given by ]V/(w+l-651360 657-42^- 1 ) 2 ; and three series having
limits 43825-00, 49012-03, 50748-26, with the variable terms given by
JV/(w-f-l-384406+446-70A~ 1 ) 2 . The number of lines in a series ranges from three
to five.
In germanium (Ge. At. wt.=72-5; At. No. = 32) Paulson finds three triplets
with separations of 1416, 903, and two pairs with a separation equal to the sum
of these.
Constant difference pairs have also been noted by Paulson among the arc lines
of titanium, with separations of 71, 64, 779, or 1166. Only a small proportion of
the lines of this element are included.
As regards the remaining elements of Group IV., carbon and silicon, some
unpublished results obtained by the writer are of interest and may be briefly
mentioned.
Carbon (C. At. wt. = 12-0 ; At. No. = 6). The arc spectrum of carbon shows only
one line, A2478, in the whole range of spectrum from the extreme red to A2000.
It is possible that there may be other arc lines in the Schumann region, but the
spectrum has not been obtained under conditions which permit them to be
distinguished as such with certainty. Observations in this region have been made
by Wolff, McLennan,|| and Millikan.^
In the spark spectrum of carbon, C + , the writer has established doublet series
for which Ar=10-8 and which have 4Af for the series constant as in other series of
ionised elements. With greater energy, a triplet is developed near A4647, having
Av=13-0, 5-5 ; this may possibly represent C++, or the second-step ionisation, but
it has not yet been possible to prove this by establishing the triplet as a member
of a system of series having $N for constant. With the greatest energy, as Merton**
* Zeit. f. Wiss. Phot., 12, 16 (1913).
t Ibid., 13, 313 (1913).
J Proc. Roy. Acad. Amsterdam, April, 1912.
Ann. d. Phys., 42, 837 (1913).
|| Proc. Roy. Soc., A. 95, 272, 327 (1919).
|j Astrophys. Jour., 52, 59 (1920).
** Proc. Roy. Soc., A. 91, 498 (1915).
M2
164 Tables of Series Lines. CHAP. xvm.
has shown, certain other lines are developed in the carbon spectrum which correspond
with lines in the spectra of Wolf-Rayet stars. The wave-lengths of the most
prominent of these are 5812 and 5801, forming a pair with Av=31-4. This may
perhaps represent a third-step ionisation, but the evidence for this is not complete.
Silicon (Si. At. wt. =28-3; At. No. = 14). Lockyer has shown that successive
spectra are developed in this element as the energy of excitation is increased, and he
has designated them Si I., Si II., Si III., and Si IV., the first representing the arc
spectrum. These observations, however, covered too restricted a range of spectrum
for the investigation of series, and have been extended by the writer.
Evidence has been obtained that the arc spectrum, Si I., includes a system of
triplets, in which Av=146, 77. There is a diffuse triplet, with a normal set of
satellites, a normal sharp triplet, certain other triplets not yet classified, and a
number of pairs having separations of 146 or 77. Series formulae have not been
calculated, but the lines in question are undoubted arc lines.
The spark spectrum, Si II., shows a system of doublets in which Av=60-0.
The principal, sharp, and diffuse series are well represented in the spectrum under
suitable conditions of experiment, and the series constant for them is definitely 4AT.
The doublets are therefore to be assigned to ionised silicon, Si + .
Additional triplets associated with that observed in Si III. by Lockyer have
been obtained, but not yet in sufficient number to allow of the definite calculation
of the series constant. It is not improbable, however, that the triplets represent
the second-step ionisation, for which the constant would be 9AT, and they may be
provisionally assigned to Si ++ .
Lockyer's Si IV. was represented by a well-known pair in the violet. The
separation Av is 164, and three other pairs with the same separation have now been
found in the ultra-violet. Further observations in the Schumann region may
establish the character of these pairs, but meanwhile it may be supposed possible
that they represent the third-step ionisation, Si + + + .
It should be noted that the separations of the doublets and triplets in the
successive spectra of silicon are related to the corresponding separations in carbon
in very close proportion to the squares of the atomic weights.
These observations of carbon and silicon are of special interest in connection
with Kossel and Sommerfeld's displacement law to which reference has already been
made (p. 74).
ELEMENTS OF GROUP V.
No series have been identified in any of the elements of this group. Constant
difference lines, however, were noted by Kayser and Runge in the spectra of arsenic,
antimony and bismuth, and in other elements of the group by Paulson. In all
cases, the constant separations refer to pairs of lines, two or more separations being
involved.
It is scarcely possible to summarise the data usefully, and reference must be
made to the original sources.*
In the case of nitrogen, it should be observed that Stark and Hardtkef have
obtained a spectrum which they have described as the arc spectrum. The more
familiar line spectrum thus becomes the probable spark spectrum, N+. Prior
to Hardtke's work, certain lines developed under a more powerful stimulus were
* Kayser's Handbuch, Vol. II., and Paulson's papers previously quoted,
t Ann. d. Phys., 56, 363 (1918.)
Elements of Groups IV. and V.
165
obtained by Lockyer, Baxandall, and Butler, and described as the enhanced lines of
nitrogen. This spectrum was afterwards described more completely by Fowler,*
and should, perhaps, be considered to represent the second-step ionisation, or N++.
The matter is of some importance in connection with Kossel and Sommerfeld's
suggestive displacement law, and the following particulars of pairs of lines may be
quoted :
A LA.
V
Ar
Probable principal pair
/ 4097-33 (10)
\4103-39 (8)
24399-30
363-22
36-08
Probable diffuse pair
(4641-91 (3)
\ 40-65 (10)
I 34-17 (8)
21536-84
542-70
572-83
5-86
35-99
Possible sharp pair ...
(4867-14 (4)
\ 58-82 (3)
20540-23
575-40
35-17
The measures of the last pair are less satisfactory than those of the first two.
* Monthly Notices R.A.S.,80. C92 (1920).
CHAPTER XIX.
GROUP VI. OXYGEN, SULPHUR, AND SELENIUM.
Our knowledge of the series lines of the elements of this sub-group is due to Runge
and Paschen.* Each element shows two distinct line spectra, which are well known
under the names of the " compound line " and " elementary line " spectra. These
names were assigned to the oxygen spectra by Schuster, f and were adopted by
Kayser for the corresponding spectra of sulphur and selenium. The two spectra are
developed respectively by uncondensed discharges and condensed discharges of
moderate intensity, and the names given by Schuster were based upon the sup-
position that complex and simplified molecular groupings were involved in the two
cases. The compound line spectra were distinguished as the " series spectra " of
the respective elements by Runge and Paschen.
When powerful condensed discharges are employed, a third system of lines is
produced in oxygenj and sulphur, ^ and it would seem convenient, provisionally,
to follow the plan adopted by Lockyer in the case of silicon, and to distinguish the
compound line, the elementary line and the third line spectra by adding I, II, III to
the chemical symbol of the element, as 0. I, O. II, O. Ill, &c. In general terms,
the three spectra have been described as the arc, spark and super-spark spectra.
Intermediate stages may be readily obtained by suitable adjustment of the gas
pressure, diameter of capillary tube and intensity of discharge.
It is only in the first line spectra that series have at present been identified.
OXYGEN.
O. At. wt. =16-00; At.No.=8.
The compound line, or O. I, spectrum of oxygen exhibits a system of narrow
triplets, and a system which was at first described as consisting of doublets. The
latter, however, were only partially resolved, and the separation was only given for
one line of the sharp series, namely :
A6046-348 (7) =vl 6534-34
A6046- 120 (2) =4 6534-97
Since most other spectra which show triplets also show singlet series, but never
doublets, it is possible that the oxygen series in question may really consist of single
lines. ] | They have been entered as singlets in the table.
The first principal triplet shows separations Ar=3-4, 2-7, according to the
* Ann. d. Phys., 61, 641 (1897) ; Astrophys. Jour., 8, 70 (1898).
t Phil. Trans., 170, 41 (1879).
J Fowler and Brooksbank, Monthly Not. R.A.S., 77, 511 (1917).
Lockyer, Proc. Roy. Soc., A. 80, 55 (1907).
II See also Sommerfeld, Ann. d. Phys., 63, 224 (1920). Recent investigations by Dr. Cataliin,
however, have suggested that the members of this system may be very narrow triplets.
Oxygen, Sulphur and Selenium.
167
measures of Runge and Paschen ; the more recent determinations by Meggers and
Kiess* are
A
V
Av
7771-928
74-138
75-433
12863-28
859-62
857-49
3-66
2-13
and the intervals are thus brought into closer accordance with the separations of
the triplets of the subordinate series. It may be recalled that the principal triplet
occurs in the solar spectrum, and has its origin in the solar atmosphere.
The limits given in the tables have been adapted from those calculated for the
sharp series by Dunz. The fundamental series should be in the infra-red, where
observations have not yet been made.
Among the unclassified lines of O. I are two " inverted " wider triplets, with mean
separations 7-7 and 12-6, which Hicksf has suggested may be united with /5037 of
the D series in a " new diffuse series," for which he has calculated the limit 22926.
A considerable number of lines of O. I have been observed in the ultra-violet by
Schniederjost.J but their relation to the established series, if any, has not yet been
traced. The wave-lengths of these lines are included in the table of unclassified lines.
In the spectrum of O. II, six pairs of lines, with separation Av=179-9,have been
identified by the writer, but the series arrangement remains to be discovered.
O "SINGLETS.
PRINCIPAL. IS mP.
15 = 33043-3.
DIFFUSE. IP mD.
!P=21207-2.
X, Int.
V
m
mP
X, Int. v
m \ mD
8446-38
11836-1
(1)
21207-2
11287-3 1 8857-2
(2)
12350-0
4368-30 (10)
22885-8
(2)
10157-5
7002-22 (4) 14277-3
(3)
6929-9
3692-44 (7)
27074-7
(3)
5968-6
5958-53 (O
o >o ^ * co
CS CO Co Co
CO W OS <* OS
CO CM l 1 '-I CO
CO CO CO CO Co
"* OS >* OS >*
Cs OS Os Oo Oo
Co OS Os OS OS
OS >O Co 10 l-l
Z Z^ Z- CO CO
Os O5 Os Os Os
co 1-1 z- CM *>
>0 >0 >* >* CO
Ob OS Os Oi OS
IO
tN GO CO O5 1C
CO i I O OO l>
p 1 t- # O I>
CO -^ CO (N O
CO OS CO CO OS
OS t~ CO 1C CO
CD CO OS CO CO
cq -H os oo i>
i 00 1C (N O
CO "* CO