mÈÉM

A

ON THE OCCURRENCE OF RARER
ELEMENTS IN THE NETHERLANDS

EAST INDIES

A SPECTROGRAPHIC INVESTIGATION
IN THE RANGE 3600-5000 A.

W. VAN TONGEREN

BIBLIOTHEEK DER
RIJKSUNIVERSITEIT

utrecht.

SI■

quot;W

''mm

V '

\ i-

; • * V ■ ^ • ♦ ^^

im

ON THE OCCURRENCE OF RARER ELEMENTS IN THE
NETHERLANDS EAST INDIES

«

iS^-'Z'Siè

24486999

ON THE OCCURRENCE OF RARER
ELEMENTS IN THE NETHERLANDS

EAST INDIES

A SPECTROGRAPHIC INVESTIGATION
IN THE RANGE 3600-5000 A.

PROEFSCHRIFT

TER VERKRIJGING VAN DEN GRAAD VAN
DOCTOR IN DE WIS- EN NATUURKUNDE AAN
DE RIJKS-UNIVERSITEIT TE UTRECHT. OP
GEZAG VAN DEN RECTOR MAGNIFICUS DR.
TH. M. V. LEEUWEN. HOOGLEEI^AR IN DE
FACULTEIT DER GENEESKUNDE, VOLGENS
BESLUIT VAN DEN SENAAT DER UNIVERSI-
TEIT TEGEN DE BEDENKINGEN VAN DE FA-
CULTEIT DER WIS- EN NATUURKUNDE TE
VERDEDIGEN OP MAANDAG 10 OCTOBER 1938,
DES NAMIDDAGS TE 4 UUR

DOOR

WILHELMINÜ8 BLANDINÜS CATHARINU8 VAN TONGEREN

GEDOKEN TE 'S-GRAVENHAGE

AMSTERDAM
D. B. CENTEN'S UITGEVERS-MAATSCHAPPIJ N.V.

1938

Deze dissertatie verschijnt tevens als de eerste en tweede aflevering van de
^^ne „Contributions to the Knowledge of the Chemical Composition of the Earth's
'-rust in the East Indian Archipelagoquot;.

BIBLIOTHEEK DER
RIJKSUNIVERSITEIT
U T R E C H Tl

'■X'èTî-ï^.

V

m.

îi-ïf]

AAN

Bij het beeindigen mijner academische studiën voel ik zeer sterk den wensch
U, Hoogleeraren in de Faculteit der Wis- en Natuurkunde van wie ik mijn
wetenschappelijke opleiding mocht ontvangen, daarvoor mijn dank te betuigen.

Hooggeleerde Schmutzer, hooggeachte Promotor, de uitgebreidheid van de
wetenschap waarin gij mijn leermeester zijt geweest en de groote vrijheid welke
ge Uwen leerhngen en medewerkers laat in de keuze van onderwerpen voor hun
studie hebben mij verschiUende malen en ook weer bij het bewerken van dit
proefschrift, doen afdwalen naar zóó ver verwijderde grensgebieden van de mine-
ralogie dat ik U des te meer dank verschuldigd ben voor den steun die gij me
steeds op onbekrompen wijze gegeven hebt voor het uitvoeren van mijn werk
en voor Uwe raadgevingen waarmee ge iemand telkens, als ongemerkt, weer een
heel eind in de goede richting weet te leiden. Voortdurend houdt ge bij ons het
besef levendig dat wat de moeite waard is om gedaan te worden ook de moeite
waard is om goed gedaan te worden en steeds bleek dit niet alleen de beste, maar
tevens de kortste weg te zijn. De jaren gedurende welke ik U als assistent ge-
diend heb behooren daardoor in wetenschappelijk opzicht tot de meest leerrijke
van mijn leven.

Hooggeleerde Rutten, de wijze van symbiose volgens welke te Utrecht
geologie en mineralogie tesamen gaan, heeft groote voordeelen en steeds meer
ben ik het gaan betreuren dat ik van de daardoor geboden mogelijkheden niet een
beter gebruik gemaakt heb. Ik acht het daarom een gelukkige omstandigheid
^t mijn toekomstige werkkring me ruimschoots gelegenheid zal geven mijn
fchade in dit opzicht in te halen. Ik ben U zeer erkentelijk voor Uw onderricht
in historische en algemeene geologie, waarvoor ge bij mij een werkelijke belang-
stelling gewekt hebt, meer echter nog voor de sfeer van zorgvuldig waarnemen
en cntisch werken waarin gij Uwe leerlingen opvoedt en welke zoo wonderwel
samengaat met en in velerlei opzicht nog gestimuleerd wordt door de prettige
Verhouding onder de geheele bevolking van Uw instituut.

Hooggeleerde Moiik, Uw met zooveel liefde gehouden voordrachten over de
oclcnikunde van Nederlandsch-Indië en in de eerste plaats de daarbij ontwik-
'elde algemeene gezichtspunten hebben mij zeer getroffen. Vooral wilde ik U
echter clanken voor het hart dat gij me onder den riem gestoken hebt in een voor
quot;^ij moeilijke periode van mijn studie.

Hooggeleerde Schoorl, zeer prettige herinneringen heb ik aan dc gastvrije
ontvangst welke mij op Uw laboratorium steeds ten deel viel. In het bijzonder
l^nk ik U echter voor Uw onderricht in de analytische chemie en voor de hulp
gij me bij mijn verder werk steeds gegeven hebt cn waarvoor men bij U niet
vergeefs aanklopt.

Hooggeleerde Vening Meinesz, Uwe colleges over geophysica hebben mij-
een ^oote belangsteUing voor het geologisch aspect dezer wetenschap gegeven.

Hooggeleerde Brouwer, Uw onderricht in de tectonische geologie heeft
mij^kenms doen maken met een zeer grootsche zijde van de geologische weten-

De banden van vriendschap welke in mijn studietijd geknoopt zijn zuUen
deze jaren onvergetelijk voor mij maken.

TABLE OF CONTENTS

list of illustrations....................xv

INTRODUCTION..............................................i

PART 1. THE SPECTROGRAPHIC DETERMINATION
OF THE ELEMENTS ACCORDING TO ARC ME-
THODS IN THE RANGE 3600—5000 A ..............5

chapter 1. apparatus and other requirements ....nbsp;7

Choice of method..............................................7

Choice of the apparatus........................................7

Description of the spectrograph..................................9

Other apparatus................................................10

Adjustment of the spectrograph................................11

Dispersion of the spectrograph..................................16

Electrodes....................................................18

Chemicals.....................................19

Use of mortars................................................21

References to chapter i........................................22

chapter ii. operating technique..........................23

General considerations..........................................23

Preparation of standard mixtures................................24

General considerations on the preparation of standard mixtures ...nbsp;25

Choice of concentration steps....................................27

Scheme for the exposures......................................27

Development of the plates......................................29

Taking the spectrograms........................................29

Cathode layer effect............................................30

References to chapter ii........................................33

chapter iii. the determination of the elements . ...nbsp;34

General considerations..........................................34

First group of periodical system..................................37

Lithium, Sodium............................................37

Potassium, nibidium.....................

Caesium....................................................39

Copper...........................

Silver...............................................

Gold..........................; •nbsp;42

Arc spectra of (H), Li, Na, K, Rb, Cs, Cu, Ag, Au................42

Second group of periodical system..............................43

Beryllium, magnesium........................................43

Calcium, strontium..................................^

Various phases in the evaporation process..............45

Barium................. 'nbsp;^^

Radium, zinc................. quot; 'nbsp;^^

Cadmium.................. ' 'nbsp;^g

Mercury................49

Arc spectra of Mg, Ca. Sr, Ba, Ra, Zn, Cd, Hg . . . . ! ^ !nbsp;50

Third group of periodical system..............' quot;nbsp;50

Boron, aluminium.............. 'nbsp;^q

Scandium................. • • •nbsp;^^

Yttrium..................................quot; '

Lanthanum.......

. .nbsp;..................... 54

Emission of spectra by rare earth metals........... 55

Cerium........................................^^

Praseodymium...................

Neodymium..................

Samarium....................................^q

Europium..........................................^ j

Gadolinium..................

Terbium ..................

Dysprosium........................

Holmium, erbium....................

Thulium, ytterbium...................'

Cassiopeium, gallium ....................

Indium, thallium................... ' ' ^g

Arc spectra of tervalent elements..............[ 59

Fourth group of periodical system..............' 70

Carbon, silicon....................................quot;

The band spectrum of carbon electrodes......................70

Titanium......................................• • . •nbsp;^^

Zirconium..................................• • • .nbsp;^^

Hafnium............................................^^

Thorium, germanium.................

Tin...........................

Lead............... 76

Arc spectra of C, Si, Ti, Zr, Hf, Th. Ge. Sn. Pb......' ' 77

Fifth group of periodical system..........

Phosphorus, vanadium.............

Niobium (Columbium)..................

Tantalum....................................................80

Arsenic, antimony, bismuth..................................81

Sixth group of periodical system................................82

Chromium..................................................82

]\Iolybdenum................................................83

Tungsten................................84

Uranium....................................................86

Seventh group of periodical system..............................86

IManganese..................................................86

Rhenium....................................................87

Arc spectra of V, VI and VII group of P.S..........................88

Eighth group of periodical system..............................88

Iron........................................................88

Cobalt, nickel................................................89

Platinum metals..............................................90

Arc spectra of Fe, Co, Ni, Ru, Rh, Pd, Os, Ir, Pt..............92

Explanation of table 1..........................................93

Table I. (table of analysis- and coincidental lines)..............95

Explanation of table II........................................110

Table II......................................................Ill

Explanation of table III........................................112

Table III. (table of ultimate and sensitive Unas).........113

References to chapter III......................................119

PART IL ON THE OCCURRENCE OF RARER ELE-
MENTS IN THE NETHERLANDS EAST INDIES .nbsp;122

CHAPTER IV. DETERMINATION OF A NUMBER OF RARER ELE-
MENTS IN ROCKS AND MINERALS OF THE EAST INDIAN

ARCHIPELAGO..............................................122

General considerations..........................................122

Choice of the samples......................122

Discussion of errors............................................123

Experimental data..............................................124

Presentation of the results......................................124

List of locahties................................................124

Table IV. (list of localities)....................................125

Results of the spectrum analysis of the East Indian samples ....nbsp;135

Calculation and accuracy of average percentages..................135

Errors in averages..............................................137

Table V. (Amounts of rare elements in East Indian samples)..........139

Discussion of the results........................................^^^

Apparent radii of atoms and ions in crystals........................161

First group of periodical system.................

Lithium, sodium, potassium, rubidium, caesium, copper, silver, goldnbsp;162

Second group of periodical system..............................163

Beryllium, magnesium, calcium, strontium, barium.......163

Zinc, cadmium.......................166

Mercury..........................167

Third group of periodical system................167

Boron, aluminium, scandium, rare earths...........167

Occurrence and spectrographic sensitivity of rare earths ...nbsp;168

Yttrium, lanthanum, cerium.................170

Praseodymium, neodymium, samarium, europium, ytterbium, other

rare earths.........................171

Calculated average percentages of rare earths.........171

Importance of further work with the samples.........172

Gallium, indium thallium..................173

Fourth group of periodical system...............173

Carbon, silicon.......................173

Titanium..........................174

Zirconium, thorium, germanium, tin, lead...........175

Fifth group of periodical system................................176

Phosphorus, vanadium, niobium, tantalum, arsenic, antimony, bis-
muth ........................................................176

Sixth group of periodical system................................17

Sulphur, chromium, molybdenum, tungsten, uranium............17

Seventh group of periodical system..............................17

Manganese, rhenium..........................................17

Eighth group of periodical system..............................17

Iron, cobalt..................................................17

Nickel, platinum metals......................................178

Average composition of earth crust in the N.E.I....................178

References to chapter IV......................................179

INDEX OF PARTS I amp; II........................................180

LIST OF ILLUSTRATIONS.

Fig- 1 — MANNKOPFF-type of spectrograph........................8

Fig. 2 — Arrangement of spectrograph and auxiliary apparatus . .nbsp;12

Fig. 3 — Electric circuit for the arc..............................12

Fig. 4a — Arrangement of first diaphragm in collimator..............13

Fig. 46 — Wrong arrangement of diaphragm in collimator............13

Fig. 5 — Dispersion-curve of the spectrograph (3600—5000 A) . . . .nbsp;18

Fig. 6 — Arc spectra of elements belonging to first group of P.S. *)nbsp;42

Fig. 7 — Arc spectra of elements belonging to second group of P.S.nbsp;50

Fig. 8 — Arc spectra of elements belonging to third group of P.S. .nbsp;69

Fig. 9 — Arc spectra of elements belonging to fourth group of P.S. .nbsp;77
Fig. 10 — Arc spectra of elements belonging to V, VI and VII Gr. of

P.S....................................................88

Fig. 11 — Arc spectra of elements belonging to eighth group of P.S..nbsp;92
Fig. 12 ^ Ultimate and sensitive lines in the arc spectrum between 3600

and 5000 A.................... • •nbsp;^^8

Fig. 13 — Apparent radii of atoms and ions in crystals according to
V. M. Goldschmidt...............

Fig. 14 __ Comparison between occurrence of lanthanides and sensitivity

of their spectrographic determination..........169

P.S. = Periodical system.

- V-^

INTRODUCTION.

In the course of the mineralogical and chemical examination of some rocks
and mmerals from various localities of the western part of the East Indian Ar-
chipelago the author's attention has been drawn to the presence of many of
ne rarer elements in these samples Furthermore, a study of the geological
^na pctrographical literature of this country, especially with regard to the oc-
currence of mineral deposits of economical interest, together with some spectro-
g aphic work on the occurrence of chromium and rare earths in tin ores perfor-
ée! in the Mineralogical-Petrographical Institute of the University of Got-
ingen m the summer of 1935 stimulated my plan to make a systematical in-
vestigation of this subject.

There are a number of reasons — many of them not being entirely free of
pportunism — which definitely lead me to undertake the work in the limita-
T^ion in which it is presented here.

^ ^ • a natural consequence of the political unity of our country with the
^ast Indies we are especially interested in questions regarding this territory.
^ u of course this reason alone would never have lead me to undertake a work
equinng so much preparation if the scientific interest of the question might be
cloubted; this, however, is not the case:

a. The geographical situation of the N.E.I. ♦) on the joining of thecircum-
staTl quot;mediterranean geosynclines has brought about the exceedingly un-
^ le position of this part of the earth's crust together with the formation of

ve'rv^nbsp;^^ ^^nbsp;volcanoes. Consequently, the archipelago has a

y interesting geological position. For a full discussion of this subject the reader
reterred to the introductory chapter of Rutten's well known work 2) on the
ogy of the N.E.I, and to Brouwer's lectures dealing with the same subject,
lead tnbsp;^^^ absence or presence in definite amounts of rarer elements may

und 1nbsp;regarding the lines of differentiation in the magma in the

dine fti!nbsp;consequently to a clearer understan-

exD^rn pétrographie units in the archipelago. Secondly, there might be
which hnbsp;corrélation with phenomena of subcrustal convection currents

Thirdl Tnbsp;probable by the gravitational work of Vening Meinesz «).

the N F t ^nbsp;investigation is of interest for a discussion on

matinnbsp;^^^ geochemical viewpoint and may also furnish useful infor-

concerning the science of geochemistry in general.

N.E.I.: Netherlands East Indies.

Some of these questions and related subjects will be dealt with in a part

to tslpa™nbsp;-y be refeS

C A thorough knowledge of the distribution of the rarer elements mav
valuable infonnation regarding to questions of economical quot;ncTaVd L

Talnbsp;occurrences of m-

d In the last decades the importance of so-called rare elements for nhnf
^owth has been established, though as a matter of fact the physiol gi a ^unc
tion of these elements is as yet in aU cases not clearly undLtood However
when more mformation is available, the conditions for further research in this
branch of science become better and the practical interest in questions of aer
culture IS already above doubt. Of course the occurrence of the required elements
m the ongmatmg materials from which the soils are built up. the rocks and the r
minerals, is of prime importance.

.. As far as is known to me a systematical investigation of the occurrence
of mmor constituents has not yet been made in some larger part of the earth's
surface and consequently this work might be of interest for a critical conside a
tion of the value of geochemical work of this kind in general

2. A few years ago this institute has been supplied with a large glass prism
s^c rograph, especially suitable for work in the blue and violet'parts oMhquot;
visible region and the near ultra violet (3600-5000 A). Included in this rani
are many important lines of most of the rarer dementi suspected to be preset
n the matenal to be examined. Consequently this subject seemed to offer an
exc Uen opportunity for the first application of this apparatus in the regular
course of mineralogical and chemical work in this institute

newLTntnbsp;^^^^dy present in the geological museums and when

new methods of investigation are going to be employed it is important - besides
being economical in obtaining the required samples - first to apply these m

valuable correlations with known properties can easily be made.

4. Almost certainly it might be expected that during this work the author

which he has devoted his student years at this university. Apart from the rea-
ons already mentioned, this seemed to offer an attractive subject for a disser-

wh?L'vA^ f ' most desirable way of expressing my gratitude to those
wno have been my teachers in these years!

The work has been carried out in the Mineralogical Department of the State-
University of Utrecht since the beginning of 1937 with varLs interruptions and
continuously since January 1938.

As already explained this is a first exploration of the available material and
claims to give rather a general aspect of the matter under discussion than aiminjr
at an - as yet - unattainable completeness. On several occasions the author
has felt an arduous desire for a quartz-spectrograph to make possible a more

acknowledgments

complete study of all the elements which might be expected in a given sample.
In accordance with the general character of the work I have not used refinements
such as rotating sectors and other useful spectro-photometrical devices; more-
over this would have meant an undesirable delay in the publication. I hope that
several of the adhering defects may be redressed in the future.

The part of the work which is now ready deals with two subjects; the spec-
trographic determination of minor constituents in silicatic samples with the arc
method in the region of 3600—5000 A is treated in the first division, the second
comprises the actual determination of a number of rarer elements in over three
hundred East Indian samples.

The spectrum negatives and the samples, together with my descriptive notes
have been deposited in the collection of the Mineralogical-Geological Institute
at Utrecht in such a way that they are easily accessible for further research.
The wavelengths are marked on many of the negatives.

I am indebted to many people in this country as well as in India and Ger-
many and I should like to express my gratitude to them:

Most of all I wish to express my sincere thanks to Prof. Dr. Ir. J. Schmutzer,
Head of the department of Mineralogy of the State University at Utrecht, for
^is kind permission to make use of the facilities of the institute and for putting
at my disposal literally everything which was required for the work and for con-
structive criticism when reading the MS., moreover for his personal interest and

encouragement.

To Prof. Dr. Dr. h.c. V. M. Goldschmidt, now at Oslo, who kindly received
me m his splendid institute at Gottingen University for some time in the summer
of 1935. I had there a unique opportunity of becoming familiar with spectro-
graphic work and I am much indebted to Drs. H. Bauer and H. Hormann, who
guided my first steps in this field of research. I shall never forget the pleasant
time spent in Gottingen with my German colleagues.

P^of. Dr. L. Rutten, Head of the Mineralogical-Geological Institute of
he State-University at Utrecht, for placing the East Indian Collection of this
mstitute at my disposal for selecting the samples required for the investigation
and for his interest in the spectrographic work.

To Miss Dr. C. E. Bleeker, Head of the „Physisch Adviesbureauquot;, Utrecht,
o undertook the construction of the spectrograph and auxiliary apparatus
m an entirely satisfactory way and who met all my demands in this regard.

To many friends and colleagues in Utrecht and elsewhere with which I had
e opportunity to discuss a number of problems encountered in my work. It is
impossible to mention them all, but I must make an exception for Dr. W. Nieu-
Wenkamp with whom I shared the assistantship at our institute during a couple
years; I shall always remember our pleasant co-operation.

To my friend R. Wardell, Utrecht, who revised the English MS. and also
aided me with the reading of the proofs.

To Mr. J. H, M. v. Dijk, Utrecht, who assisted me in preparing the drawings
tor figures 6—12.

REFERENCES

') W. v. Tongeren — Chemische Analyses van Gesteenten van Poeloe Berhala
Proc. Acad. Sci. Amsterdam. 38, (1935), 634.

- Mineralogical and chemical Composition of the Syenitegranite from Boekit

Batoe near Palembang, Sumatra. Netherlands East Indies. Proc. Acad Sci Amster
dam. 39. (1936). 670.nbsp;' ^mster-

r •nbsp;~nbsp;over de Geologie van Nederlandsch Oost-

Indie. Gronmgen. Den Haag. 1927.

T quot;quot;'Tt'nbsp;Netherlands East Indies. New York

London, The Hague, 1925.

*) W. v. Tongeren — reference 1. p. 670 note 1

in the F^rth ^^^^^^^^nbsp;quot; Gravity and the Hypoihesis of Convection-Currents

m the Earth. Proc. Acad. Sci. Amsterdam, 37. (1934), 37.

PART. I.

the spectrographic determination of the elements
according to arc methods in the range
3600—5000 a.

.:. -,nbsp;St- : quot; . Ti- : ' -i

■■ •■; ï
....... .......üp-iß^: 1

. i

'lip -PÄiÄPfc^'Ä^^?;

CHAPTER 1.
APPARATUS AND OTHER REQUIREMENTS.

choice of method.

Only the spectrographic method seemed to meet the demands of a simul-
taneous investigation of so many different elements. The reasons are: 1. though
emg very specific as to the results, the procedure of spectrographic analysis
iffers but slightly for the various elements and generally one working-scheme
can be adapted to these different demands; 2. the time required for a complete
spectrographic estimation is much less than for a chemical determination of even
a much smaller number of elements. Also, the spectrographic procedure almost
entirely avoids the troubles otherwise frequently encountered in procuring the
required chemicals of such a grade that they are entirely free from traces of ele-
ments to be determined or at least contain only such small amounts as are ne-
ghgible; 3. in many cases the spectrographic methods allow much smaller concen-
trations to be estimated than is possible with chemical methods; 4. if need be,
the analysis can be performed with a minute quantity of the material to be exa-
mined, this being of importance for the investigation of minerals which often
cannot be selected in large quantities without extreme difficulty.

On the other hand, the small amounts used for spectrographical work form
more or less a serious drawback in cases where only large masses may be sup-
posed to represent the true composition of the sample. This is the case with rocks,
ut often a representative sample can be obtained without working with extreme
quantities and in other cases it is generally possible to overcome the difficulty
y separately examining the mineralogical components of a complex sample.

Consequently, and as an exhaustive investigation was not intended, I
ave applied only the spectrographic method instead of a combination of this
procedure with chemical and other physical methods. It is to be hoped that this
metnod will give enough data for a preliminary survey.

Choice of the apparatus.

miun^^'fTnbsp;the analysis of samples in which elements like iron, chro-

of wh- ^nbsp;quot;'^amum, zirconium and rare earths might be expected, all

red Fnbsp;complicate spectra, a large dispersion apparatus was requi-

larao 7nbsp;reasons we decided to begin a tentative investigation with a

large glass spectrograph.

°nbsp;apparatus

In the first place, such an apparatus is easier to handle than an ultraviolet
equipment and less expensive.

Secondly, in its more limited spectral range it is superior in brightness as
well as in dispersion compared with grating, respectively uviol glass and quartz
instruments, while this range would contain enough important lines of the ele-
ments believed to be encountered in the N.E.I.

During my stay in Göttingen I had the opportunity to see and appreciate
the remarkable stability and convenience of operation of a special type of glass
pnsm spectrograph, constructed by Dr. R. Mannkopff, Head of the Physical
Section of the Göttingen mineralogical laboratory i). The construction of a spec-
trograph according to these principles was undertaken for us by the Physisch
Adviesbureauquot; of Dr. C. E. Bleeker, Utrecht. The arrangement of the instru-
ment is shown in fig. I.

T = Total-reflection prism
C = Condenser
P == Camera

In order of succession the optical parts are:

An achromatic collimator lens, aperture 61 mm, focal distance F. = 1500
mm, (R).

Two equilateral prisms of the following dimensions:

1.nbsp;height of prism 60 mm, length of face 102 mm, (Di).

2.nbsp;height of prism 65 mm, length of face 110 mm, (Dg).

(owing to the dispersion in the first prism the pencil has been broadened
when passing through the second one)

A right-angled total reflection prism with two quadratic surfaces, the length
of the cathetus being 70 mm, (T).nbsp;» ^ gtu

description of the spectrographnbsp;9

A planoconvex single flintlens, diameter 80 mm, focal distance Fd = 1500 mm
(for sodium light), as objective, (0).

The required optical parts in light flint glass were ordered from Messrs.
B. Halle, Nachf., Berlin. Though uviol glass is of course more transparent in
the near ultraviolet, this has not been chosen as material for the prisms as the
mentioned advantageous property is counteracted by the consequence of the
smaller dispersion, thus necessitating the introduction of more prisms and there-
fore causing an increased absorption and reflection.

As well-known, the objective is purposely not chromatically corrected, the
chromatic aberration giving rise to an inclined position of the photographic
plate, which has the additional advantage of increasing the lenght of the spec-
trum without increasing the focal distance of the spectrograph.

With these parts an experimental model was made to check the calculations
or the properties of these special lenses and prisms. As the results agreed per-
ectly with our expectations, the construction was carried out and the complete
spectrograph installed to be adjusted in a quiet room of the institute.

The spectrographic laboratory is situated opposite the dark room and has
a wmdow facing W, so that scarcely ever direct sunlight enters. Consequently,
emperature fluctuations are not extreme and are in fact small enough to be
unobnoxious for the correct adjustment of the apparatus.

Description of the Spectrograph. Laboratory number i. 218.

The essential parts are built on a platform supported by a frame of two
heavy and rigid double T-beams, connected at both ends by two sohd steel pla-

rPQ T*! *nbsp;X

• ine instrument rests with three points on metal shoes, placed on two so-
lid tables.

Resting at one end on a stable aluminium platform are the three prisms,
acn on a separate turntable which is fastened in the required position by a
screw. Each prism is held in a fixed position by three rectangular metal pieces
and only very slightly pressed on the table (to avoid irregular double-refraction
owmg to internal tensions) by screws ending in circular cork-covered metal plates.

At the same end of the spectrograph are the two lenses in circular cells. They
h e adjusted along the axis of the spectrograph by a rack and pinion motion,
accessible from the exterior, viz. when the cover of the instrument is in position.

slits and the camera, likewise mounted on a heavy steel plate, are at
he opposite end of the apparatus. There are four slils of fixed width, viz. 0.6,
• , 0.04 and 0.02 mm, mounted on a circular plate which can be revolved round
a horizontal axis, parallel, to the optical axis of the collimator. The widest slit
permits easy working in the case of several adjustment operations where rather
a great intensity is required as will be mentioned further on.

The effective length of the slits can be varied by diaphragms in front of
em i have made several lenghts with aluminium plates which can be conve-
th^^ ^ fixed. The plate holder can be turned round a vertical axis going through
e middle of the camera and slides in a vertical direction in two grooves being

auxiliary apparatus

operated with a knob. There is a scale at one side and a counterpoise ensures
an easy running. Colhmator and camera are separated over the whole length

llh^'^T Tl'rnbsp;^^nbsp;by a metal hood

which protects the mtenor against dust and stray-light. A smaU removable cover

above the prism-tables aUows their easy adjustment.

factory^ construction is aU-metal and proved to be well-made and entirely satis-
Other Apparatus.

At the slit-end of the spectrograph a clamp was attached bearing a verti
cal rod with another clamp to take a condensing lens. An ordinary magnifvine

Senser'quot;'' ^^nbsp;F, = 100 mm) was used as a s^eLJ

990 v^n rnbsp;resistance for taking the required current from the

2.20 v D.C. mams have hkewise been designed and executed by the Phy^^isptt

Adviesbureauquot; of Dr. C. E. Bleeker. Utrecht.nbsp;quot;

The fl.c stand is designed so as to permit a convenient regulation of the
runmng ^rc. The whole arc can be moved in a horizontal plane by two levers
the one for bringing the arc into the optical axis of the collimator, the other -
working m a direction perpendicular to the former _ for the movement zn this
axis^ The height of the arc is regulated by a rack and pinion motion. At any
fixed height of the centre of the arc (e.g. in the optical axis of the spectrograph)
the length c^n symmetncally be varied by another rack and pinion motion Thi
works extremely light as the weight of those parts bearing the anode is counter-

to aXtth; rrnbsp;' P-vision has been made

to adjust the relative position of the anode over the cathode, the anode being

the upper electrode. Electrode holders with spring clips were used throughout the

work and permit the changing of electrodes in a minimum amount of time tL

used and still hot electrodes being firmly gripped with metal forceps and rem;4d

to rnake place for the new ones which can be inserted by hand. Of course most

of the metal pieces of the arc stand are insulated to avoid as much as p;squot;b

the coming into contact of the operator with parts under electric tension For

ItlTZ'^fT ^nbsp;''nbsp;th^ to indicate

at a glance if the arc stand is under tension.

The resistance is of a very heavy type. When placed in series with a nor-
mal arc, taking 50-60 V from the 220 V mains, the difference between two stages

is approximately 0.25 A so that any desired current between 2 and 10 A can be
regulated within narrow limits.nbsp;a can oe

For several reasons the arc lamp is placed under an exhaust hood connec-
ted with stove-pipe to an electric fan placed outside in the open air. This is nec L
sary for keepmg the surroundings of the arc clean and to rLove the obnoxTous
and sometimes poisonous vapours which are liberated by most arcs from the labo
atory air. Moreover the large amounts of rare elements sometimes evapL'^ed
m the arc would contammate the atmosphere so severely as to make impossible

adjustment of the spectrographnbsp;11

any correct determination of the element in question for some time. Also, the
arc is effectively protected against draughts so that it bums in a more constant
way, which is of vital importance in the case of arcs between metallic electrodes.
The arc itself can be observed through a window in the hood. This window can
be made of dark filter glass and should protect the eye also against ultraviolet
radiation. A convenient material is at hand in any spectrographic laboratory
in the form of overexposed photographic plates (two, three or more as required).
The image of the arc on the slit — because of the strong illumination of the lat-
ter is observed through such a plate as this gives a uniform, yet effective,
weakening of all the wavelengths, so that the true colour of the emitted radiation
can be observed, which is often of much importance. This glass is conveniently
placed between the operator and the slit and may be supplied with a lens of sui-
table focal distance to yield a magnified image of the slit and surroundings.

A plane stainless steel mirror cemented to this glass plate (which is placed
Usnbsp;^^^^ ^^^ optical axis and which can be moved horizontally in

s own plane) permits the use of another apparatus to check the emitted radia-
on, e.g. a small spectrograph for the entire visible region. This is much more
convenient than making changes in the correct alignment of the whole optical
ram, nothing being changed now in the relative position of the different parts,
of dquot;^^*^ ^^ disposal a small sfectrografh for the visible spectrum, composed
a irect vision microspectroscope according to Abbe combined with a suitable
camera (Messrs C. Zeiss, Jena; 128130/1284 and 128135/102). As the working
aperture and the focal distance are small, the required exposures are much longer
an with the large model and consequently I have used it only now and then
or photographic work, though it has more often proved a valuable help for the
visual inspection of a spectrum with a lens of six-fold magnification placed at
the end of the camera.

des ^quot;^^^quot;t consumed by the arc and the tension between the two electro-
des can be read from two moving coil instruments (A and V), placed on a marble
Jise at the camera-end of the spectrograph. All the instruments are arranged
ha dr ^^^^^ ^^^ operator so as to permit their simultaneous control and
^an uig by a skilled person. This of course is only possible with — and another
^^vantage of -- Littrow- or mannkopff-type spectrographs. A sketch of the
^^nbsp;electrical circuit a schema is represented

(See page 12)

Adjustment of the Spectrograph.

he correct place of arc and centre of the condenser in the optical axis of
tne collimator was secured in the following way:

The widest slit was used and the condenser removed. The beam of light in
ne spectrograph could easily be followed and directed on the middle of the colli-
lattnbsp;^^ adjusting the arc stand. Without changing the position of the

er, the condenser was put in place in its clamp so that the arc's image was formed

adjustment of the spectrograph

s

WB

Fig. 2. Arrangementofspectograph and auxiliary apparatus

WB = Workbenchnbsp;O = Operator's place '

M = Mam switchnbsp;a = A-meter

AS = Arc standnbsp;S = spectograph

R = Resistancenbsp;y = V-meter

Fig. 3. Electric circuit for the arc

a and V ^nbsp;^ mains

a and V = Moving coil instrumentsnbsp;r ^ . ,

N = Neon signal lampnbsp;=

o; the optica. Lin

of the arc sharply on the slitnbsp;^ ^ ^ focussing the image

-'withstanding the

mator. By reflections on Tenbsp;^Hi-

through the optical den,ents;irquot;g„ a Ma

result. It is true that these wa^ls are patat d dl n i unbsp;plate may

that almost any nat surface of ^hif d . ,nbsp;hiquot;

rather perfect mirror, provided the angle of incidence i f

h.ht was removed hy applying two diphri^Mw^iT^^^^^^^^^^

adjustment of the spectrograph

is placed on half the focal distance and is of such ovaloid form that the light of
any part of the slit can freely reach any point of the collimator lens, but effectively
protects the lens against reflected light from the walls (fig. 4a). There remains,

choice of the region

aperture of the spectrograph. Luckily, the same result is obtained without the
mentioned disadvantage by placing a second circular diaphragm a short distance
jay from the coUimator lens. Both diaphragms were connected with the walls
of the collimator by means of black paper and are painted dull-black

The focal distance of the achromatic collimator has been checked and nrn-
ved to have the value of 150 cm as claimed by the makers. So the dÏtancquot;
between the slit and the centre of the lens was given this length with t^e aM

OI 3, rUiG,

The prisms were placed in the position of minimum deviation for a strong
Ime m the violet part of the spectrum of ordinary, impure carbon electrodes
which later on - accordmg to a sketch made of this spectrum - proved tn Ù
the Fe-Line at 4307.906 A. In the adjustment chosen for the spectrograph thi
line IS approximately m the middle of the range.

Therefore, the operating room being darkened, with the w-ide slit and onlv
the first pnsm put m plaee, a speetnim was formed on a sheet of white paper
by plaeing a speetacle glass of 2 Dioptries behind the prism. By turn „rthe
atter, he reqmred position was easily found, then fixed by the sere^To^ t

rnbsp;quot;quot;quot;nbsp;-»d devTaTlèn

a ter ,his had been replaeed on its table. Then the third, total-reflecHng S
and the eap of the spectrograph were replaeed. The openings between f âme worT
and cap were treated with dark wax. The prisms ean stin J rllh V?,, u
coverable opening in the cap. Placing the rS^g L f oU, rniltquot;®^ '
tical position was a tiresome procedure, no special adiu!w„ ^ I quot;quot;quot;

however, thenbsp;'n Sene'ro.,

especially when band spectra are formednbsp;'' P'''^f'=rable.

3/0. The first number

choice of the regionnbsp;15

number the ultimate lines. Of course the use of these numbers is liable to some
objections, mainly because the region above 5000 A is as yet only rarely used for
spectroanalytical purposes, except for visual work and some special occasions
where even the infrared has been used The same tendency, however, can be
remarked in Kayser's well-known tables *) where the above mentioned regions
take respectively 48, 49, 46, 35, 25 and 25 columns. Anyway, these figures are
only given as an example without claiming an absolute value for them, but as
far as my experience reaches, they fairly demonstrate the real conditions and
furthermore the large number between 3500 and 4500 may give some justification
for the use of a special glass prism spectrograph for the work in this region.
Moreover, the lines of the rare earths are not taken into account in the above
figures; as they fall mainly in the mentioned range, the numbers are even more
favourable than is shown here. More than two thirds of the lines of the rare
earths given in Gerlach and Riedl's tables (being 73/20 lines) lie within the
range of our spectrograph.

3. The sensitiveness of normal photographic emulsions is likewise decreasing
enormously for wavelengths higher than 5500 A and they still work rather slowly
in the range 5500—5000 A. The use of ordinary panchromatic plates only partially
overcomes this difficulty as was found with an e.xperiment using the small Zeiss-
spectrograph mentioned previously. Also, the use of spccial emulsions for general
^vork is liable to several objections which need not be discusscd here.

On the other hand, going too far to the short side would cause a rather
severe restriction of the wavelengths near 5000 A, where valuable lines of other-
wise unattainable elements would be cut out. This is due to the unhnear disper-
sion of prism-spectrographs, however, an example will make this dearer:
^^ In the setting as finally chosen, the range of 3600—5000 A is taken on a
^^ X 24 cm plate. Going down to 3500 A would mean ending at the other side
approximately 4570 A and consequently exclude among others, strong lines
13a. Cd, Eu, Li. Ni. Re, Sr, Zn and Zr.
^ esides, this would have resulted in a slight decrease of dispersion over the
0 e range, as the necessary inclination of the camera to the optical axis be-
toT^nnbsp;(together with the degree of curvature of the photographic plate

its °nbsp;course is an advantage). But the construction in

present state does not allow a smaller inclination of the camera than the one
quu-ed for the range 3600—5000 A; 3600 A is the lower limit, fixed by the
eciianical possibility of the instrument.

taknbsp;tledded that 5000 A would be the highest limit of the region to be

account, firstly as the working-conditions for the spectrograph arc
linesnbsp;this region, secondly as there is a large number of important

also^ ? .^^any dements, finally as some other reasons, though being unessential,

bourhood 360oquot;nbsp;^nbsp;^^ expected in the neigh-

the di!^^nbsp;of the total-reflecting prism, required to throw the 5000 A in

irection of the end of the camera was found by visual observation of the

dispersion of the spectrograph

t'uZhf quot;nbsp;been found the third

turntable was equally fixed.

tionnbsp;relative^sition of camera and objective (distance and inclina-

tion of the camera) and the curvature of the plate holder to suit the focal curve
of the objective, a number of exposures was made with a fixed position of the
camera and at an increasing distance between photographic plate and objective
In a number of consecutive exposures, sharply represented regions grLuallv
movmg m the direction of the dispersion could be detected. ThI platfyS
the required information to calculate the correct curvature and inclinaL of
the camera and the position of the lens. After these had been carried out a few
more exposures were still necessary to give the finishing touch to the adjus'tmen^

Dispersion of the Spectrograph.

The disperdon of the spectrograph has been measured in a spectrogram
taken with an xron arc, the values for the round numbers being i^aTed
between the wavelengths of the nearest iron lines. These wavelenrth. r k
found by comparison with the well-known reproduction^^
available from Messrs. Adam Hilger Ltd London and hZlT
Kavsfr's tables , For various purposes it i; aS;

spec rograms of the iron arc resulting from ..or. or /..s prolonged exposur^
For the purpose now under consideration, it is best to use a spectmm ofnot n
great density as the position of the lines is not accuratelv LVn T I 1
are too heavy; besides the smaller number of hn squot;^^

DISPERSION OF THE SPECTROGRAPH

17

Whil fquot;-nbsp;columns the differences for 50 respectively 100 A are given,

is m tquot; ^^^^ ^^^^ column the dispersion in A/mm, as calculated from these data,
nentioncd. Fig. 5 is a diagram of these values. The accuracy of these measure-

the v^nbsp;enough, though being not very great, to give a sufficient idea of

variation m dispersion in the range.

cynbsp;^^^^ ^ comparator which was not available, a better accura-

micr°quot; ^^ obtained.-except with a very time-consuming procedure, viz.
causT^^^'^ measurement with a micrometric ocular. This, liowever would have
c a cumulation of errors, which has been avoided in the above measurements.

accessories

Fig. 5. Dispersion-curve of the spectograph.
Accessories.

Trr^'T-nbsp;^^^ ^^^ ^l^^trodes to contain the

r/ hnbsp;the chemicals of sufficient purity to p!

pare the standard mixtures and to dilute the samples. Further the mortars used
m the preparation of the sample i).nbsp;mortars used

Electrodes.

As electrodes specially purified carbon rods have almost exclusivelv bonr,
used; m some cases electrodes of pure metals and nf rnr.T r

oTr —

long each and of 5 mm thickness has been obtaiL'd from M ss 1 P TA L?
The Hagne and after purification was reserved for specialircritof work ° '
For the greater part of the work, carbons supplied bv Messb, rT.^:

(Nürnberg,, Kinomarke Noh.s Hs'Sialtlr^rm^ wequot; tXI'

piJ

ching to which the comparison negati^s arc iL^L w

will have to be made, as the Dutch manXcW ,,nbsp;»•»■■gt;• quot;'w standards

.eems to be incapable of producing t^mt the , Tnbsp;quot;quot;quot;

nature of the emulsion if ot no Ipo Snce wlthi^H ' quot;f' quot;

consiLfd'o/:: -iTaVTrcr^^^^^nbsp;.he plates

electrodes and chemicalsnbsp;'nbsp;19

These carbons were rather pure and in this regard not much inferior to many
speciahy purified carbons which are on the market, but they still contain large
amounts of calcium and especially alkalies. The rods were broken m pieces of
33-4cm with the aid of pincers. With some experience one gets enough skill to
break them almost perpendicular to their axis. These ends have been treated in
resistance glass cylinders for some weeks with aqua regia, prepared from specially
pure acids (pro analyse) of Messrs E. Merck, Darmstadt. The further solution
of impurities was obtained by leaching them during 3—4 months with hydro-
chloric acid of gradually decreased concentration and finally by a treatment with
shghtly acidihed distiUed water. The reaction was purposely kept acid, as then
^e glass is hkely to be less attacked than with a neutral or even alkaline medium.
Ahe rods were dried over a micro-burner in porcelain dishes which had been
previously treated with aqua regia and steam. They were stored in clean card-
oard boxes and the core was taken out shortly before use over a length of 4 mm
^ith the aid of a dentists drill. The diameter of these holes was approximately
mm and they contained 9 ± 1 mg of normal rock samples when hlled up
itn the aid of a blunt steel pin, obtained by removing the worked part of a drill
^ used for making the bores.

For easy handling, a wooden holder with 8 or more holes to take the elec-
rocles after these have been filled should be prepared.

titi ^ ^^^ purification the carbon electrodes still contain considerable quan-
^ les of calcium, aluminium and iron (e.g. in spectrograms 379 and 379 bis or
Consequently, they were of no use for the determination of these elements,
in t? ^ ^nbsp;disadvantage for our work as explained in the third chapter

Crandnbsp;elements. Other impurities are: Sr, Ba, Ti,

eleme t •nbsp;»mounts arc so low as not to affect the determination of these

therequot;nbsp;quot;quot;quot; samples which were investigated and consequently

nine thnbsp;^quot;»quot;tJ^cr purifying the electrodes. Particulars concer-

recLinnbsp;elements will be found in the third chapter; general remarks

Göttin 1?nbsp;lanthanum as in the

occurrfquot; ,nbsp;chmcnt has caused many difficulties. The possible

scnmmr''nbsp;elements in carbon has been explained by V. M. Gold-

and Cl. Peters in an interesting communication on this subject»).

Chemicals.

rally ofnbsp;investigation were all of sufficient purity, gene-

of thenbsp;^quot;•'^quot;ty from dependable manufacturers. To obtain the spectra

cornpoumr^quot;/!; ?'^^^quot;^quot;^^ ^^quot; ^^ necessary more spectrograms taken with

are reserri fnbsp;'^»^e been consulted. The chemicals referred to

tute- thpnbsp;tquot;gt;s kind of work in the spectrographic laboratory of the insti-

numbers are stated in my note-books so that a special

certaintv m ^^nbsp;spectrum can be traced with absolute

compoundsnbsp;quot;quot;quot; mixtures derived from these pure

for atnbsp;IInbsp;^^^ standard mixtures of which enough is present

^^dst Ö0—60 exposures.

20nbsp;base substances for standard mixtures

Impurities in the substances used for the preparation of the standard mix-
tures, provided they are not excessive, are without influence on the results. This
however, does not hold for the compounds used as a base, as even relatively small
impurities may be considerably above the hmit of the spectrographic sensitivity
for the concerned element as shown in table 11 of the third chapter. As explained
in the next chapter, only quartz powder and sodium carbonate have been exten-
sively used in the preparation of the base. The sodium carbonate used for our
experiments was a pure preparation of Merck as commonly used in silicate rock
analysis for the decomposition of insoluble samples and no objection whatever
could be found against using this substance after extensive trial in exposures
using different samples of carbon electrodes with or without addition of silica
This also avoids Böse's objection to the comparison of ordinary exposures with
spectrograms taken with the electrodes alone «). Böse supposes many of the con
taminations being present in the form of carbides, which are difficultly volatile
and only yield the metallic components in a state in which their spectra are
emitted after decomposition by the molten sample. Of course, this is no supnort
in favour of Boss's supposition and personaUy I believe the temperature in an
arc consuming nearly 10 A high enough to decompose all these substances

The quartz taken as base for the standard mixtures was obtained from
Merck m the form of coarse particles, washed with hydrochloric acid The im
punties amount to 0.08 «/o as has been determined by volatilizing the ;ilica Z-
sent in 2 g of the sample with hydrofluoric acid to which a drop of sulphuric Lid
was added to retain titanium. Consequently, the amount of 0.08 % is too hiX
as ^ ineludes the impurities in the appreciable quantity of hydrofluoric acid whfch
had to be taken and which proved to be of the same order of magnitude The
impurities seemed to be essentially iron and titanium which has been clLk d
by taking spectrograms of the scrapings from the platinum crucible in wl e
operation had been carried out. (Spectrogram 292). As explained this do no
give us an Idea of the actual amount of impurities in the siL and the onTy pTs

tages of TiO, than 0.003 %. (Spectrogram 336)nbsp;® ®

tars quot;iffri^rn b quot;nbsp;iquot; -quot;or.

eoohng. Many small fissures are caused bvTs trL/r!™^nbsp;^^

energy required to reduce the powder to the ts red fTnl T

most agreeable way. Nevertheless it is saf^l L fT , quot; ''«quot;'^^ed ,n a

or ether. If either of these linnlt a v '' Pulverization under alcohol
contain inorglnfe quot;m^t'eT quot;nbsp;^^ quot;-y 0° -t

use of steel mortarsnbsp;21

Use of Mortars.

The contamination of the rock samples by the crushing in steel mortars pro-
ved to be negligible on account of the following consideration. The mortar used
for this work caused a contamination of 0.08 % of FeaOg on crushing quartz from
coarsely crystalHne fragments to dimensions of less than 0.3 mm') A spectro-
gram taken with the pestle (618 A) revealed the absence of Cr (which is extre-
mely sensitive) and the presence of manganese in a quantity of probably a few
tenths of a percent. Even if it amounts to one percent it will not be found in
greater quantity in the sample than corresponds with the manganese content
of the other requirements for the exposure. As the presence of iron is quite irrele-
vant for our investigation — for the reasons which are given in the third chapter,
paragraph dealing with this element — and as no other elements were detected,
I did not hesitate to use this old diamond mortar which materially abridged the
time required for crushing a sufficient quantity of the rock samples, as compared
With exclusively using an agate mortar. Mortars of more recent manufacture
should be very carefully tested in the described way, as however trustworthy
hey may be in other regards, their use in spectrographic investigations is not at
all unobjectionable owing to their often high content of elements which in this
respect are to be considered as troublesome impurities. In connection with this
question it may be interesting to know that one of my colleagues working with
a so-called „Abreiszbogenquot; using grooved copper electrodes to take the sample,
regulariy observed percentages of chromium and vanadium which could not be
etected by other operating techniques and moreover were not to be expected
m the samples in as high amounts as seemed to be present. In the end it turned
out that the knife used to rub the powder in the grooves was of a very good quality

..Spezialstahlquot;!nbsp;b m a

REFERENCES.

') R. Mannkopff — Ueber eine Bauart von Prismenspektrographen mit lanfr^r
Brennweite. Z. Physik, 72, (1931), 569.nbsp;^

Teil T^hnbsp;fquot;nbsp;Emissionsspektralanalyse III

leil. Tabellen zur qualitativen Analyse, Leipzig, 1936.nbsp;. ^^^^

') V. M. Goldschmidt, H. Bauer und H. Witte — Zur Geochemie rl.r am
metalle II. Nachr. Ges. Wiss. Göttingen, Math.-Physik. KlasJenbsp;39

Folge, Band l. Nr. 4).nbsp;'

L. W. Strock — Zur Geochemie des Lithiums. Nachr Ges Wi^«:
Math.-Physik. Klasse. (1936), 171. (Neue Folge. Band l, n; 15)nbsp;^«quot;»quot;gen,

') H. Kayser — Tabelle der Hauptlinien der Liniensoektra T7i
Wellenlänge geordnet, Berlin. 1926.nbsp;'quot;'«quot;Spektra aller Elemente nach

») V. M. Goldschmidt und Cl. Peters — Ueber die

•) R. Böse — Optische und Spektrographische Untersuchnn.rnn ^
Jahrb. 70A (1936). 538.nbsp;Untersuchungen etc.. Neues

') W. v. Tongeren - Gravimetrie Analysis, p. 34. Amsterdam, 1937.

CHAPTER II.
OPERATING TECHNIQUE.

General Considerations.

The principles of spectrum analysis have been described in a more or less
extensive way in several recent publications. Here, exclusively particulars con-
cernmg the special operating technique of this investigation will be dealt with.

Apart from Geissler tubes and the like, mainly three light sources may' be
considered for the purpose of exciting spectra; these are:

the flame,
the electric arc,
the electric spark.

For the examination of non-conducting materials, such as rocks and many
mmerals, the arc is especially suitable, the energy being high enough to excite
^^le spectra of most elements in sufficient intensity. The same cannot be said
^^ the flame wherein only a limited number of elements produce their spectra,
^ven when an oxygen-acetylene burner is used. The prominent advantage]
^owever, of the arc method over both flame and spark excitation, is that the
duT^ tT^^ ^^ treated according to more or less complicated chemical proce-
^ res. These otherwise are required when non-conducting or insoluble substances

rest^T'^'^iquot;^^'nbsp;throughout the case witii our samples. This not only

import fnbsp;required for the analysis, but. which is much more

freoinbsp;possibility of contaminating the sample with the impurities

quently occurring in the reagents even of analytical quality.

Kopff^^^T'^'^' ^^^^ ^^^ methods, the cathode layer effcct. discovered by Mann-
there V^-nbsp;^^ exploited to increase the sensitivity 10—100 times,

the c •nbsp;the other methods. The cathode layer effect causes

cannbsp;the available atoms in the place where the maximum energy

Henc^ ^quot;PPlit-^t^. namely in the region of the large potential fall near the cathode.
Point^of Tnbsp;fulfilled for an intense radiation originating from this

^^ the s ^nbsp;increased concentration and higher excitation,

of thenbsp;is observed as an enhancement of the lines when an image

gives a^'^^ ^^ ^^^^rply focussed on the slit of the spectrograph, which instrument
slit Annbsp;conforming to the intensity differences at the place of the

of excitât- advantage of the use of the cathode layer effect is, that the conditions
quantitat'°quot; are much more stable in the region concerned, consequently, for the
ive estimation, results found with this method are superior to tlios

preparation of standard mixtures

othenvise obtained as they are more dependable and go down to lower concentra

tions. The same is reached for a number of elements which are volatiHzed ahreadv

at low temperature, by adding a sufficient amount of alkalies (e.g in the form

of sodium carbonate) to the sample, which also increases the intensity of the emit

ted radiation for many other elements. This is not yet entirely explained, but as

the procedure has been extensively used in this work, a possible explanation wiU

be ^ven m the next sections whilst an extensive analysis of the evaporation urocess
IS also communicated there.nbsp;process

Preparation of Standard Mixtures.

The elements which had to be considered were divided in groups of ten mem
bers whose sensitive lines do not coincide. These groups are:

IV.nbsp;V.

Knbsp;Cu

Rbnbsp;Sm

Csnbsp;Eu

Mg Gd
Canbsp;Tb

Scnbsp;Dy

Ynbsp;Er

Sinbsp;Yb

Nbnbsp;Hf

Monbsp;Re

Later on, a coincidence which had better be avoiHpH t ^ .
tween two lines of eerium and neodymium as win te dt uL^n ^ ne Vf '
Apart from this coincidence - which does notnbsp;« . ^

second group - the combinations ^^«Ttot S Siquot;
course many others are likely to be as suitabira^rL?.^ T
in special e^es, e. g. when a more limited numbe of eÏmentLTbe ,

It need not be said that this procedure greatly Xes t^
and time required to prepare the standard .„ V

the method of taking onLriesorsUtro^ll fCquot;^^^^^ quot;

ever, inadvisable further to extendThÎpSquot; quot;»w-
dences in the higher steps will materialCctle

thor^ghly mixed in

for this work as it is of the utmost imnort^nr. f k? 'nbsp;quot; ^^ ^^l^en

before a further step is made. One t ntCro' hifr^quot;^^ ' ^--ogeneous mixture

of the pure quartz powder which had p'rnbsp;^^^^^ to 900 mg

procedure mentioned in the former chapter and nnbsp;according to the

tamed by grinding the powders for a quLter of ^n ^nbsp;^^

4 01 an hour m the agate mortar under

preparation of standard mixturesnbsp;25

alcohol. These mixtures, containing 1% of each of the oxides concerned, are the
starting-points for the further dilutions. These have been prepared by weighing
out the calculated amounts required for the composition and thoroughly mbdng
them in an agate mortar as described in the first step.

General Considerations on the Preparation of Standard Mixtures.

For preparing the standard mixtures it has been recommended on several
occasions i) to make use of compounds in which the elements to be determined
occur combined in the state, in which they are expected to be present in the sam-
ples to be examined. Only then, the conditions in the arc are supposed to be
exactly the same for standard mixture and sample to be investigated; e. g. it is
without doubt that a number of elements are evaporated from artificial mixtures
m an eariier phase of the whole process than from the natural compounds, a fact
which is especially remarkable in the case of vanadium. There are, however, other
reasons which plead for the use of the oxides. As mentioned furtheron, no difficulty
Will be encountered in preparing even very dilute mixtures when the artificial
oxides (or other compounds) are used. Down to the lowest concentrations,
ne mixtures are entirely homogeneous, a fact which obviously finds its explana-
lon in the relative softness of these materials and the fine state of division in
Which they generally are supplied, as compared with the natural compounds. In
e second place, the special operating technique employed in this investigation,
amely the dilution of the samples with half their weight of sodium carbonate,
causes, prior to the evaporation of the sample — but for the most volatile com-
pounds — a homogeneous melt in which the sample is entirely decomposed. That
^^ e most volatile substances are not retained in either case is of course an advantage,
tend'quot;^^*^^^ ^^^^ conditions for the emission of their spectrum lines similar, besides
Con separate them in the spectrograms from the accompanying elements.
^^ ^sequently, both these reasons enable us to have a great tolerance in this regard,
esti^^î-^^^^quot;^quot; comparison of the intensities is the base of the quantitative
in t'T ' ^P^^'t ^rom the fact that the real state of combination of the elements
silicatnbsp;investigated is unknown in a number of cases, e. g. with

mino ^^^^^ ^^ ^^ possible that some elements are concentrated in the
there^^^^ evenly distributed in the essential components,
of th^ •nbsp;^^ con.siderable influence on the promptitude

of j^J^'^.^^^P^i'ation from the sample. Moreover one can dispense with the necessity

^King or procuring a chemical analysis of these natural materials,
for th ^^nbsp;whether the simple procedure of grinding the components

a hom^ ^^^quot;^^rd mixture in a mortar for some length of time effectively yields
mixture down to the lowest concentrations. Considering the minute
centrât^^^ involved this is indeed vcry remarkable; in the lowest con-
nig thi ^^^^ ^^^ ^^^^ standard mixtures in my work and for a sample of 10
^hichnbsp;^^^^^ 0.00001 mg of the oxide occurs in the sample, an amount

atoms ^f thnbsp;unimaginably small, but nevertheless contains a number of

0 the order of magnitude of 10quot;—10quot;. which neither has any meaning for

^^nbsp;sensitivity of spectrographic method

us. Even then there is no reason to assume the distribution over these 10 me
should not be uniform, or that the extinction of the spectrum hnes is due to the
mhomogemty of the sample, rather than gradually becoming too weak for
photographic or even visual observation. The very fact that, even in the lowest
concentrations, the spectrum lines gradually fade away is to my idea the strongest
support for the homogenity of the samples as other^vise irregularities would no
fail to be produced. There is, however, stiU more evidence m favour of this opinion
In the higher concentration steps, the particles of the oxides, or the mixed
oxides, are easily visible among the clear quartz-particles when a suspension
of the mixture is microscopically examined. At intermediate concentrations
when the homogenity of the sample is beyond doubt, they entirely disappear I
mchned to suppose that, owing to the action of the sharp-edged quartz fr^^L .
the added substances are evenly distributed in a very^thin kyrovequot;
available surface of the quartz powder. The number^of quartz parfc^^^^^
m 10 mg (in the higher concentrations also the other fragmentsWnfstin
the total surface after due grinding also is very reassuring in th s reLTd Anl Tquot;
the actual samples are concerned, there seems to be no difficultyS ^ quot;quot;
a sample which in every 10 mg is representative for the larger Lmp ' e.n C
m view of the fact that a great number of rare elements occTun^^^
hidden (German: quot;getamtquot;; Goldschmidt) in minemTs of Z ^^ ?
with which they share some properties, in the kTXœ the Z
their atoms, respectively ions, in crystal lattices

I cannot resist mentioning the extreme smallness of the auintit,-.c u- i

can be detected of many elements by spectrum analys s and vvh

from a calculation resembling much the reLning of B^sL Ld W

their classical and for the spectroscopist of to-day stiU veryTnteS^^ quot;
tion on spectroscopy 2).nbsp;j- ciy inceresting investiga-

Once the place of a spectrum line being known it s padlv ^k
twinkling of an eyequot;, which may well be the thousand h n' ? ^.
over which the radiation emitted by the ImLtTv1 . ''
is volatilized in the arc. This corresponds vlhl ! ' .nbsp;^

the element if a concentration of 0 0001 «/

a quantity of 0.000 000 0! ms of

quite a large „„.ber of elements this istn'Les ^ateftll'Ir

a. an actual test independable) sodiun, fiame-roactanbsp;Tquot;

chemical test methods, however, amounts can h.nbsp;quot;quot;'•«■■n bio-

favourably with the spectro-analytical m l^d in tW quot;quot;'quot;l

activity of 50.10'AE/g (avena-units per , .nbsp;'«s an
detected by a deviatol of 1quot; oflhrste^Jd t !quot;quot;quot; -V

to a sensitivity of 0.000 000 002 me and mor» „nbsp;corresponds

quite recent time, the extreme W? whlh cln t 7nbsp;In

viz. molecular proportions, have been dataS to .nbsp;'quot;^stance,

for a molecule of uncommon high molecular weight ''nbsp;quot;'»quot;Sl'

concentration steps; exposuresnbsp;27

Choice of Concentration Steps.

In the tables occurring in this work the various concentrations steps which
have been mentioned are: 1%, 3.10-i%.nbsp;etc. down to 10-quot;%, though

the percentages in the standard samples slightly differ from these amounts in
the intermediate steps between two entire powers of ten. In reality the steps
between the concentrations in consecutive samples are all the same, which is of
course a great advantage as well from the point of comparing the intensities as
for the preparation of the samples. Whilst the first point scarcely needs elucidation
(formerly, often the steps 1/0.5/0.1 etc. were used, which has the disadvantage
of irregular decreasing intensities) the latter may be explained in a few words.

In preparing the next lower concentration starting from a given step, the
calculated amount of the latter is weighed out and enough of the base is added to
yield the required dilution. When quantities of a gram are prepared, an amount
of 0.3162 g is necessary for the dilution. Consequently, I undertook the making of
an aluminium weight of this mass, but had the good luck to break a piece of
alummmm from a larger plate which proved to have just the required mass
Without necessitating further work.

A quantity of 0.6838 g remains for taking spectrograms of the concentra-
lon steps, except for the last where 1.00 g is available. In every concentration

g was thorougly mixed with 0.15 g of sodium carbonate, whilst the rest is
jailable as such. All these standard mixtures are kept in small corked glass-
u es and are in the collection of the spectrographic laboratory of our institute
quot;ey may easily be traced with the aid of my note-books.

Scheme for the Exposures.
the ^^^^^ extensive trial the following scheme has been adopted for preparing
e spectrograms of the East Indian samples and the comparison spectra:

1.nbsp;ca. 30 seconds with 2—3A *)

2.nbsp;30 seconds with 5 A

3.nbsp;30 seconds with 7 A
4—7. 4 X 30 seconds with 9 A

the d'nbsp;exposures the plate holder was given a displacement in

linesnbsp;perpendicular to the dispersion of twice the length of the spectrum

movinbsp;^^^^ spectrograms can easily be compared. The time required for

if necquot;- ^^^nbsp;optical axis, changing the position of the plate holder,

is apir^'^-'^^ flccreasing the resistance in the circuit and readjustment of the arc
second^ seconds; thus limiting the time of the actual exposure to 25
disapn^'nbsp;seconds the hollow part of the cathode has entirely

quot;P and 'nbsp;^^^ ^ ^^^^ ^^nbsp;substance with which it was filled

Volatilenbsp;intensity of spectrum lines belonging even to the least

^^ not fQnbsp;'s perceptible. Consequently, the period was sufficiently long,

quot;lation^'^ ^vholly evaporating the sami)le, at least to yield all the necessary infor-

^Vith undiluted rocks this may slightly vary according to the alkali-percentage.

scheme for the exposures

Further, the above mentioned scheme has been chosen to avoid as much as
possible coincidences of lines belonging respectively to easily and difficultly
volatile substances; the low current in the beginning gives a correspondingly low
temperature in the arc, thus preventing the emission of lines which are only dif-
ficultly excited. This is also promoted by adding sodium carbonate to the sample
as a high percentage of easUy ionised substance causes a decrease of the resistance
and consequently of the potential difference and available energy in the arc

Moreover, it enabled me to dispense with the practice of giving the cathode
a special shape, as in the beginning of the evaporation process the electricity
transport is almost entirely assured by the high amount of sodium in the arc gas
column, which gives a quiet bum. Further on, owing to the gradual oxidation of
the cathode, the form is sharp enough to prevent excessive wandering of the arc
which generaUy is moving slowly along the wall of the hole in the cathode

Under these conditions the light intensity proved to be too strong even for
the not extremely sensitive phtographic plates which have been utihzed (400
H amp; D). Of course this might have been redressed by taking much shorter exposu
res either m the ordinary way or by placing a quick-rotating sector of adaotablp
aperture in front of the slit. Both procedures were rejected as it is more favoLble
to uniformly weaken the radiation than to restrict the length of the period during
which It IS a^^^owed to act on the photographic emulsion, as a much better average
intensity will be obtamed in the first case. Placing a filter somewhere in the way
of the rays is objectionable on account of the uneven weakening which is than to
be feared for radiations of different wavelengths. The best solutiL of thi proWem
IS obtamed by decreasing the working aperture of the prisms to the oper^fng

Tenfr Th' quot; TT'nbsp;^^ Photographic plate in the otTl timf

o?ZtlZ ^hnbsp;^^^ ^^^ ^^ditional advantage

of improving the image of the spectrum lines on the photographic plate nrnviZ

the aperture-restriction is cairied out by decreasing the'effSt.rort e

pnsms, preventmg the formation of slightly curved and at th^rnbsp;,

spectrum lines, which otherwise may be obsLed as\ rot

way taken by part of the pencil through ^ prTsm trlnbsp;''

fraction by a small aperture, the finaf result ofT ^: consequence of dif-

phragm is the formadon of addition

observed on both sides of the place where trLlnlV^ -frquot;quot;nbsp;^^^

tion perpendicular to the dispersion, th s fequot;^^^^^^

quality of the lines. Experiments with various d

on spectrogram nr. 71. One of these diaphrrJ ^ ^quot;^'

placed in front of the colhmator whit^ttTh 'nbsp;' ^ '

the spectrographic department.nbsp;'nbsp;reference in

It will be clear that the total length of the exnn.nmc • , . •
required for the complete evaporatfon o the samT T.'
the length of the individual exposures is set i; tTequot;^^^^
average intensity on the photographic plate for vvLv^ ^ registering a true

30 seconds seemed to be adequate The need o - rnbsp;approximately

4nbsp;need of working m the steep part of the

securing a uniform developmentnbsp;29

blackening curve of the photographic plate restricts the variation in time which
can be allowed between the individual exposures and generally a uniform period
proved to be most satisfactory. The first phase of the evaporation process — during
which the entire amount of easily volatile elements is evaporated — being ter-
minated approximately 30 seconds after igniting the arc, the length of the whole
process has been distributed over 7 exposures of 30 seconds each. As a consequen-
ce of the dimensions of sht and photographic material, two entire exposures could
be registered on one plate which consequently, takes both the spectra of one sam-
ple, the first being taken with, the other without the addition of sodium carbonate.
This determines the aperture of the diaphragm to be used in accordance with
initial light intensity, transmission in the optical train and the properties of photo-
graphic plate and developer.

It is advantageous, though not strictly necessary, to dispose of a slow-ro-
tating sector which can be placed in front of the sht and automatically secures the
correct timing of the exposures, whilst in the intervals during which no hght
can reach the slit, the necessary change in the position of the plate holder and
the regulation of the electrical current can be carried out.

Development of the Plates.

In the dark room the plate is immersed in the developer and in the beginning
of the process which than takes place, a fresh plate may be inserted in the plate
holder after the number of the exposure to be made has been marked on the
sensitive layer. In due time the experimenter's attention should be directed on
le development. As it is quite impossible always to secure entirely reproducible
conditions of temperature, concentration etc., I have thought it advisible to
^epend on the following criterion to determine whether the plate should be
jansferred to the fixing bath: at a given moment, which is quite obvious to the
^ye, the band spectrum changes of aspect and seems to become continous and
to ] . ^^^^nbsp;length. Tliis moment turned out to be a good criterion

obtain a reproducible blackening of the band spectrum and as far as the inten-
^ ICS are compared by visual inspection, no greater error is introduced by this
thenbsp;^^ ^^^^ comparison itself. Now, the intensity of the band spectrum at

sam^r^ the exposure is almost independent of the original composition of the
if at^ 1 ^^ nearly evaporated and only difficult volatile elements are present
tru Consequently, the moment in which the separate lines of the band spec-
for tlnbsp;marks the same conditions of development

^ ^ le other components of the spectrum and when this method is followed, even

obtain^dquot;^^^'quot;quot;nbsp;deviations in the conditions of the light-emission is

Taking the Spectrograms.
chantnbsp;arrangement of the apparatus is used as described in the first

^n im^'^' ^^^^^^ spherical condenser between light source and spectrograph,

^ge of the arc is formed on the slit which for the special conditions of our

30nbsp;taking the spectrograms

investigation was slightly enlarged. Full benefit of the cathode layer effect is
obtained only when the correct place of the image on the slit has been found and
maintained during the entire process. This is reached by visually observing the
spectrum after removal of the plate holder, either with the aid of a lens whose
focal plane should then be in the focal surface of the objective, or, which is
even more convenient, by simply accomodating the eye; this, however, is only
possible when the focal distance of the objective is large enough and when a
sufficient distance from the focal plane can be taken. It will be seen that the cor-
rect adjustment of the image on the sht is reached in a position in which ap-
parently the cathode is already projected there. This position being known,
— it seems shghtly to vary with the different phases of the process — no serious
difficulties are encountered in keeping the correct adjustment during the whole
period of the exposures. It is better that a small continuous spectrum of the
cathode is seen in the spectra, than that the distance of the cathode from the
end of the slit is taken too large.

Several phases in the evaporation process can be discriminated between the
moment of igniting the arc and the completion of the volatilization. Rather a
characteristical change is perceptible when a substantial part of the alkalies
has been driven off, whilst a more gradual transition exists between the remaining
two stages, which as a whole, however, show a marked difference in character.
In the intermediate phase nearly all the components of the sample are present
though of course they are continuously being evaporated one after the other.'
In the last phase are concentrated a number of difficult volatilizable elements
such as the rare earths, zirconium etc., all of which generally occur in small
amounts in samples of common type. Owing to this fact and because of the small
amount of these elements which is present in the arc gas column, even when they
are supplied in sufficient quantity, a large part of the electricity transport in
the arc is assured by the arc gas itself, a fact which has already found passing
reference in the former paragraph. The later the phase, the stronger the intensity
of the band spectrum which is emitted, while it may be neariy entirely absent in
the hrst exposure when enough sodium carbonate is present in the sample.

Cathode layer effect.

As a consequence of the high temperature, part of the amount of metallic
elements whjch is present m the arc, occurs in the ionised state. The potential
gradient in the arc causes an enrichment of these ions in a distinct layer near the
cathode. Of course this concentrât on will be greater when the substance is vola-
flized from the cathode instead of from the anode, which practice was formerly
in common use as only then the migration of the metals in the arc as a cause
of the evaporation ,s effectively repressed; it is easier to keep th mtogther tTan
to have to assemble them in the surroundings of the cathode. This cathode Wr
effect has been extensively examined by R. Mannkopff ») and in the consWe aWe
number of investigations m which it has been used to Obtain a Ire squot;

cathode layer effectnbsp;3j

method of analysis, the emission of the lines from the region of this layer proved
to be of a remarkable constancy.

In the publications of the Gottingen laboratory a slight objection is noti-
ceable against the use of larger quantities of substance than 1—3 mg and also
against the supply of additional alkali as a decreased intensity of the cathode
layer effect is feared in consequence of these procedures. There is, however, a
marked increase in intensity for a great number of spectrum lines which do not
require a high temperature for their excitation, especially as a cause of the
repressed ionisation of elements of high ionisation potential when much of an
element of a lower ionisation potential is added. On the other hand there is no
difference in conditions between an exposure of the undiluted sample and an
exposure of the sample diluted with sodium carbonate after the alkalies have
been evaporated in the arc, at least there is no unfavourable difference as will
be seen. Constant working conditions, however, are obtained during the first
phase, where after an addition of alkali carbonate the often varying amounts
of alkalies in the natural samples cannot produce different conditions in the
exposures of samples belonging to various types, which also is of the highest
importance for the standard mixtures of necessary purity which are easily
procured when simple compounds can be used as a base.

A marked increase of intensity of a number of lines which appear at all or
in full brightness only after — and often a considerable time after — the alkalies
have completely evaporated from the sample, is observed with numerous elements,
e.g. many rare earths, zirconium and others. The same is perceptible for other
elements which volatilize in an eariier phase, but this is less remarkable as a
direct influence of the addition of alkali carbonate may be supposed. The only
possible explanation for this fact seems to be that a better decomposition is
obtained in the presence of alkali carbonate and the increased intensity is due
^ at least to some degree — to the decreased amount of spluttering after this
decomposation. Even with pure iron oxide no difficulties arc encountered in
peeping the sample in the bore of the cathode, whilst otherwise it is frequently
lost as a whole, not only by spluttering.

The larger amounts of sample used in my work (ca. 9 mg when undiluted and
apart from the additional alkali 6 mg in the other case) were not detrimental
o the sensitivity of the determination, as in the eariicst phase of the evapora-
tion process the intensity of lines in the spectra of volatile elements is materially
jncreased and as in the last phase of the process the quantity of the sample has
een reduced to minute amounts by the evaporation of the quantitatively most
iniportant elements. As a consequence of the fact that the quot;interestingquot; elements
rom a geochemical as well as a spectrographical point of view emit their spectra
ler in early, or in late phases of the evaporation process, makes it unobjection-
^ le to take somewhat larger quantities for the samples and to dilute thorn with
^o lum carbonate. The influence on the intensity of the lines by adding this salt
® the sample will be mentioned for every element in the description of the
spectrographic determinations in the next chapter.

32nbsp;cathode layer effect

A disadvantageous influence could only be observed with elements such as
zinc, which are less sensitive in general, besides requiring specially deep bores
of the cathode, as otherwise they are volatihzed before they have duly emitted
their spectra. Of course these special provisions could not be made in our work.

Further particulars of general interest will occasionally be found in the
discussions of the next chapter; these are clearly indicated in the table of contents
as well as in the index.

references to chapter iinbsp;33

REFERENCES

n, . ?,nbsp;Goldschmidt. H. Bauer und H. Witte - Zur Geochemie der Alkali-

39. (Neue Folge - Band 1. Nr 4).

171nbsp;Geochemie des Lithiums. Nachr. Ges. Wiss. Göttingen. (1936).

171. (Neue Folge — Band 1. Nr. 15). p. 174.nbsp;^ ''

tuJl^Anbsp;R. Bunsen - Chemische Analyse durch Spectralbeobach-

tungen. Ann. Physik (Poggend.). 110, (1860). 161. p 168

Hilfenbsp;- Ueber quantitative Spektralanalyse mit

rtilfe der negativen Ghmmschicht im Lichtbogen. Z. Physik 70 (1931) 444

Physfk. 7Mr932r396~nbsp;lonenbewegu^g im Lichtbogen. Z.

2.nbsp;Elektronentemperaturin frei brennenden Lichtbögen.

CHAPTER HE
THE DETERMINATION OF THE ELEMENTS.

GENERAL CONSIDERATIONS

In this section particulars will be given concerning the spectrographic
determination of the elements in the range 3600—5000 A. Remarks of a more
general nature occasionally will be found in paragraphs dealing with an element
which clearly shows the phenomenon to be described, though it may be in this
regard a representative for a whole group. In some cases this proved to be a better
procedure than giving these remarks the place where they might be expected,
namely in the former section. But there, the examples would have troubled the
general trend of thought, apart from the necessity of repeating the same tings
in this chapter.

The subjects of these general remarks are mentioned in the table of con-
tents in the beginning of the book as well as in the concluding index; conse-
quently it wiU not be difficult to find the information on these themes.

The elements are treated in the order of succession of the periodical system
and are mentioned also when their quantitative determination has either not been
attempted or proved to be impossible and even when there are no lines of the
element in the region under consideration. An exception has been made for a
number of, mostly metalloid, elements (e.g. nitrogen, oxygen, halogens, etc.),
which wiU not be expected in a work on quantitative spectrography according
to arc methods. Further the table of contents and the index should be consulted.

Ah these elements have been treated according to the same scheme. First
the hnes of the element are given, often only a selection of the strongest lines.
The wavelengths and intensities stated with these lines are Kayser's values m'
and represent respectively intensity in arc, spark and Geissler tube. If in either
of these sources no emission takes place, this is indicated by a horizontal line
in the table of wavelengths for the concerned element. The spark intensities
are useful to show the character of the line and are therefore mentioned
here. When the hne is not recorded by Kayser and nevertheless had to be given
in these tables, I have tried to record the arc intensities in accordance with the
intensities of the other lines constituting the spectrum. Outstanding strong lines
are printed m heavy type. This table is followed by a scheme representing the in-
tensity of the hnes when the percentage of the element is stepwise decreased.
Particulars follow on the character of the hnes, the influence of the presence of

numbers indicating intensitiesnbsp;35

Other elements (especiaUy alkalies!) and the period of the fractional distiUation
from the arc during which the element is volatilized and consequently the emis-
sion of its lines may be observed. As the schemes applies to the arc spectrum
only the arc intensity is stated in this case. Finally some remarks are given when
the behaviour of the element in question gives rise to speculations of a general
character. Particulars concerning the operating technique and comparable items
regarding the determination of all the elements are exclusively dealt with in
the former chapter.

Concerning the numerical values indicating the relative intensities of the
spectrum lines some explanation as to their significance seems desirable. As far
as possible the same value indicates the same absolute intensity for lines of every
element. In the concentration steps where a line is not visible, the absence is
marked by a horizontal line on this place in the intensity scheme of the concerned
element. A doubtful line and also a line whose presence at some definite con-
centration is not quite dependable is indicated as (1). The number 1 is given to
the faintest line which is just visible; in the case of enhanced lines this is also the
Shortest line. Generally a magnifying glass is required to observe these lines,
especially when strong lines are in the neighbourhood. The same holds for the
next step indicated by 2. With growing intensities the values 3, 4, etc. are attri-
buted. the steps being evenly distributed until the intensity 10 is reached which
corresponds with the intensity of the stronger lines of the calcium spectrum in
tins region (3933.670 A. 3968.457 A and 4226.728 A) when a sample of an average
Igneous rock is spectrographed.

For many elements the lowest quantity which can be detennined by this
method is 0.0001 % and in any case, this has been the lowest concentration used
this work. As. under normal conditions, the difference in intensity between
|e steps IS expressed by a unit of the scale, the intensity of a line starting with
^e value 1 at a concentration of 0.0001 % will rcach the number 9 at a concen-
retnbsp;^^ «course, lines which appear at higher concentrations do not

1 0/ ^nbsp;^nbsp;^e^l^er they arc, the lower the number they get for

ce / . ^ system gives some correlation between the intensity-number at a con-
^ntration of 1. o/^ and the arc intensity in Kayser's tables, at least for lines
ose intensities gradually increase with growing concentrations. When the
^^^^cnsity increases slower, some steps are marked by the same intensities, in
gennbsp;numbers arc passed over. Fortunately, lines of the latter type

can^ •nbsp;intensity 10 as defined above and their intensities

^^ 1 easily be expressed in terms of this scheme. The scale meets the demands
'quot;^^stigation and seems to be advantageously applicable in this kind of
verv' t ^^^^^ ^^ preferable to such vague and subjective qualifications as:
onnbsp;strong, intermediate, weak and very weak, as much depends here

m.i personal appreciation of the experimenter and the type of apparatus

^ncl other utensils he uses.nbsp;yP 11

schenbsp;(I) is given with the lines appearing in the intensity

mes accompanied by the lines which are likely to be coincidental with them.

influence of coincidences

I think the value of this table has been increased by the addition of the intensities
of the disturbing lines. This has been done in the same manner as for the analysis
lines. Of course the arrangement of this table is according to wavelengths.

When looking for coincidences, the „surroundingsquot; of the line to be taken
into account vary in extension with the part of the spectrum under considera-
tion. With our spectrograph, at 3600 A two lines are clearly separated if their
wavelengths differ more than approximately 0.15 A, at the other end of the spec-
trum this amounts to ca. 0.6 A. For intermediate wavelengths the correspon-
ding values have been calculated from the dispersion curve of the instrument
For the computation of table I, I have taken the limits three times greater, but
I have not strictly adhered to them and especially I have not mentioned those
Unes of rare elements which would not be found in the intensity scheme of this
table, though obviously they will be visible if their elements are present in quan-
tities higher than 1.0 %. It is to be understood that the table applies exclusively
to quite normal cases.

As in geochemical publications the concentration is often stated as the
amount of grams per metric ton (identical with the amount of mg per kg), in
stead of the percentage, both forms are compared in the following table :

At the end of this chapter the composition is described of a powder showing

the ultimate hnes of the elements under consideration. Owing to lack of time I

could not take the actual experiment. It will be interesting to investigate whether

all the predicted hnes are really present in a spectrum taken with a powder of

this composition A table of these lines is given and their relative position is in-
dicated m fig, 12.

It scarcely needs mention that this idea is inspired by the well known Raies
eTeSIc Co quot;nbsp;Laboratories of the Gene-

lithium, sodiumnbsp;37

FIRST GROUP OF PERIODICAL SYSTEM.

Lithium

When pure lithium carbonate is volatilized in the arc the foUowing lines
are registered on the photographic plate:

4 — (Spectrograms 35 and 409)

not naZfnbsp;^^^^ ^^^^ -tremely diffuse and therefore

not particularly suitable for spectrum analysis. With dilutions of lithium car-

tho flVquot;-quot;^quot; fnbsp;^^ith sodium carbonate

loilowing results have been obtained (Spectrograms 450—459).

has evnbsp;intensity only after a substantial part of the sodium

theless tr'quot;quot;^ approximately 30-60 seconds after igniting the arc. Never-
carbon f K ^^^ stronger, especially in the lower concentrations if enough sodium

lines o 1 unbsp;^^nbsp;^^^^nbsp;Of the other

charactquot; f!nbsp;vaguely perceptible with 1. o/«. Owing to their diffuse

ments rnbsp;^'quot;cs are not likely to be confused with lines of other ele-

of this 1 ^^nbsp;^^nbsp;presence of much chromium, because a line

by Kayser^I)quot;^ coincides with the 4602.0 lithium line; this line is not mentioned

Sodium.

carir^y?^^ sodium salts in the arc the following lines or rather nebulous
^quot;catures of lines are shown:

38nbsp;sodium, potassium, rubidium

Not listed by Kayser 2).

— (Spectrograms 38 and 613)

These fogs are entirely unsuitable for spectrographical work and are only
mentioned here as they regularly appear in the spectra of mixtures with sodium
carbonate. Their fading away indicates the end of the first stage in the process
of evaporation. So these lines may be a guide at the inspection of negatives as
the lines of most elements increase considerably in intensity when the presence
of much alcali vapour no longer lowers the temperature and potential difference
in the arc.

The chemical determination of sodium in rocks according to the method of
Lawrence Smith is easy; moreover this element generally occurs in quantities
of more than 1 % of NajO. Consequently there is no need for a spectrographic
method of estimation. Many sodium determinations are already available for
the East Indies.

Potassium.

In this range of the spectrum potassium is represented by one doublet:
4044.16 K I lOR lOR —
4047.22 K I lOR lOR — (Spectrogram 430)

These hnes quickly disappear with decreasing concentrations; with 0.1 %
KjO in the mixture they are extremely vague in the first exposure of 30 seconds,
with smaller percentages they are no more visible. So they may be considered
as an appreciable qualitative reaction — the more so as visual inspection is quite
easy after removal of the plate holder — otherwise their value is rather limited.
Besides, the same as has been said about the chemical determination of sodium
also holds in this case.

Rubidium.

The spectrum of rubidium taken with our spectrograph consists of two lines
in the violet:

420L81 Rb I 8R 7R _

4215.58 Rb I 7R 5R (Spectrograms 91 and 431)

rubidium, caesiumnbsp;39

Both these Unes occupy a disadvantageous place unless special provisions
are made, which cannot be done in general. The line 4201.81 coincides with
a Mn-line (4201.767 A; Goldschmidt. Bauer und Witte p. 39) and lies near
the Fe-line at 4202.033 A ; the line 4215.58 coincides with a strong Sr-line (4215.515
Sr ir lOR), besides falling in the head of a CN-band.

These handicaps can be avoided by separating the alkali sulphates from the
rock samples (Goldschmidt. Berman. Hauptmann und Peters 2), p. 236) and
by taking the spectrum with copper electrodes. As an alternative the infrared
spectra may be employed (7800.30 Kb I 1 OR and 7947.63 Kb I lOR). but it is
clear that either of these possibilities falls beyond the limit of our investigation.

The violet lines are equally unfavourable from a point of sensitiveness;
the arc sensitivity of Kb is between 2 and 3, but concentrations of 0.1 % Rb^O
in common rocks and minerals are very rare.

The line at 4201.81 A can be seen down to 0.3 % Rb^O in spectra of sub-
stances which do not contain much iron or manganese. Intensity for 1 % between
4 and 3, for 0.3 % approximately 1.

Considering the relative intensities of the lines:

the following remarks can be made:

The fourth line is certainly absent when the first cannot be seen; the sixth
ine is weaker than the iron-lines nrs 2 and 3. Consequently, if at 4202 A a line
^s observed stronger than the iron-lines 4063.600 and 4071.743 and at the same

Mn 4035.730 is absent, the conclusion seems justified that some rubidium
present in the sample, enough to overcompensate the difference in intensity
etween the iron-lines. On this basis I believe to have found this element in some
of the samples.

Considerations of this kind can often be used with good advantage. The
arguments will not always be given as detailed as in this example.

Caesium.

Caesium lines in the range 3600—5000 A are:

3876.6 Cs Inbsp;2--

3888.6 Cs Inbsp;4 1 —

4555.3 Cs Inbsp;lOR 4 —

4593.2 Cs Inbsp;lOR 3 — (Spectrograms 103 and 432)

caesium, copper

Another strong Cs-Une mentioned by Kayser (4603.8 Cs 10 10 —) has not
been found.

With caesium the same difficulties arise as in the case of rubidium. The arc
sensitivity is equally low, the most sensitive hne (at 4555.3 A) coincides with a
rather strong line of titanium (4555,494 Ti I 9) whilst the average percentage
of caesium in the earth's crust is much lower than that of rubidium. With the
true raies ultimes of caesium, much better results have been obtained (Gold-
schmidt c.s. 2), p. 43, 53 down to 0.0005 % Cs^O using: 8521.15 Cs I lOR and
8943.60 Cs I 6R).

The hne 4555.3 Cs I lOR has an intensity of 2—3 when the mixture contains
1 % CsjO. For 0.3 % the intensity is 1 and at lower concentrations the line disap-
pears.

Copper.

In the spectrum of an arc between two copper electrodes the strongest hnes
in the range of our spectrograph are:

The conditions in the carbon arc, however, are extremely unfavourable for
the emission of the copper spectrum. In the spectrum of an arc between carbon
electrodes with a fairly heavy electrolytic copper deposit it was rather difficult
to detect even the strongest copper lines (4022.70 A and 4062.75 A) The same
has been observed in the spectrum of a carbon arc wherein a few centigrams of
copper oxide were volatilized (Spectrograms 136 and 443). More sensitive lines
are in the green region of the visible spectrum (e.g. 5218.21 Cu 10 10 —) and
especially in the ultraviolet partquot;).

It need not be said that copper cannot be estimated spectrographicaUy in
the spectrum between 3600 and 5000 A. The reason that neverthLss the lenis

copper, silvernbsp;41

in this range have been given is, that copper electrodes may now and then be
used with great advantage when carbon electrodes of sufficient purity with re-
gard to some elements can not be procured or when lines masked by too heavy
parts of the band spectrum are to be investigated. The CN bands are extremely
weak in the copper arc as in this case the only source of carbon is the small amount
of carbon dioxide in the atmosphere.

Silver.

A rather great number of lines belonging to this element are found in the
range of our spectrograph, but unfortunately neither of them is strong enough
to be registered when quantities decidedly below 1. % are concerned. So, these
hnes are only of interest when impurities in metallic silver or in argentiferous
aUoys are investigated in this range of the spectrum. As 1 had the opportunity
to take spectra with silver of the highest purity (in use as a standard for the silver
ritration at the Dutch Mint) and as not all the silver lines found are reported
in Kayser's tables I mention those lines which could be definitely accredited
to this element:

3623.
3640.
3682.3
3709.
3719.
3810.

3840.8
3908.
3915.
3968.

3981.6
4055.25
4174.
4185.

4210.7
4212.01
4311.05
4476.07
4556.0

4615.9
4668.54
4677.9
4848.15
4880.

nitrnbsp;^^ 4055.25 A and 4210.7 A arc visible when 1 % of silver as silver

•quot;ate is arced in a base of quartz powder (spectrogram 471).

roll / '■cmarkable that this soft metal takes up so much iron when being

.—^^J_o^betwecn steel cylinders. The intensity of the iron lines is greatly redu-
p

I ^ m our spectrograms this line is much stronger than the 4212.01 line.

silver, gold

ced by etching the surface layer with nitric acid. As contamination with material
from the tools is always to be feared, this precaution should never be neglected
and the amount of working reduced to the possible minimum.

Gold.

With the chloride of the metal a spectrum has been taken in which the follo-
wing lines were found:

3897.89 Au
4040.95 Au 2
4065.08 Au I 6
4437.29 Au 4
4792.62 Au 8
4811.61 Au I 3

The 4792.62 line seems to be rather persistent as I found it in a silver
sample for which a purity of 999/1000 was claimed, whilst at any rate the pre-
sence of much gold in this sample is highly improbable (Spectrogram nr. 17,
silver from Messrs H. Drijfhout amp; Zoon, Amsterdam. Bij intermediary of
Dr. j. W. a. v. Hengel, Dutch Mint).

I have not investigated further the quantitative spectrography of gold as I
do not expect to find it in the Indian samples in such quantities as can be deter-
mined without preliminary concentration.

Sm

SEfflF;

B^BByseam

3EB

Fig. 6. Arc spectra of elements belonging to first

beryllium, magnesium

SECOND GROUP OF PERIODICAL SYSTEM.

Beryllium.

Only one line of beryllium appears in the arc spectrum in the range under
consideration and this line is too weak to be of use in spectrum analysis:

4572.69 Be I 8 I — (Spectrogram 75).

In the ultraviolet very low concentrations of BeO, down to 0.001 % can
be determined (2348.62 Be I lOR *).

Magnesium.

The following hnes of magnesium have been found in the spectrum emitted
by pure magnesium oxide:

(Spectrogram 443)

Though magnesium is usually present in such concentrations in rocks and in
many minerals that it can chemically be determined, a table with the intensities
P these lines is given as information concerning the percentage to be expected
m the chemical analysis is useful especially in the case of this element.

^ The volatilization of magnesium becomes important after most of the al-
f ICS has gone and in the presence of enough of the metal this process is cons-
^^ ^^^^ »quot;tense green light being emitted (lines at 5167.33, 5172.68 and
cal •nbsp;^ substantial part of the magnesium has been evaporated,

^^^^^nbsp;colour the flame, orange and again green respectively,

ronbsp;of magnesium is rather quiet whereas iron generally splutters vigo-

th^nbsp;the possibility to discern between the two and to decide whether

end of the whole process is approaching.

44nbsp;magnesium, calcium, strontium

The hnes of magnesium are somewhat better visible when enough sodium
carbonate has been added, the band spectrum of CN being more or less repressed.

In the ultraviolet the arc sensitivity of magnesium is higher than 3, namely
between 5 and 6 for the lines at 2795.540, 2802.712 and 2852.130 A.

Calcium.

Over thirty hnes of this element lie in the range 3600—5000 A, some fifteen
of these remain visible when the concentration of CaO does not exceed 10 %.
Most of these lines are found in every negative as they are very persistent and
as it is difficult to get and keep electrodes free from calcium; even if this is the
case (f.e. after the arc has burnt for some lenght of time between clean electrodes)
the atmospheric dust contains enough of the element inevitably to contaminate
the arc flame.

These persistent lines are hsted below, for the position of the other lines the
figure nr. 7 should be consulted.

(Spectrograms 51,434 and 480).

For three reasons a table of intensities will not be given: There is no need
for a spectrographical method, the chemical method of analysis being both
simple and accurate; it would also be rather difficult to work with such a table
as the incalculable effect of the contamination interferes and as the intensity
of the spectrum lines is only slightly dependent on the percentage of calcium
in the sample.

Calcium volatihzes in an intermediate stage of the whole process after the
alkalies and magnesium, but before most of the iron and especially the refrac-
tory oxides. During this stage the colour of the flame is preponderating orange red.

Strontium.

There are at least fifty hnes of strontium lying between 3600 and 5000 A,

strontiumnbsp;45

the position of the stronger among them will be found in fig. 7. Only three lines
are important for the spectrographic estimation of the element:

4077.714 Sr II lOR lOR —
4215.515 Sr II lOR lOR —

4607.342 Sr I lOR 6 — (Spectrograms 92 and 408)

The arc sensitivity is fairly high and the intensity of the emitted radiation
duly changes with the concentration:

The vapour tension of the oxide is low, consequently the lines appear in a
far advanced phase of the evaporation process, besides only then the tempera-
ture is high enough and the energy available required for the radiation of the
strontium ion. An exception is to be made for the atom line 4607.342 which is
already emitted at a much lower temperature, viz. directly after striking the arc
during the evaporation of the alkalies. In this stage the vapour pressure is so
low that even small amounts of strontium are sufficient to maintain the arc
saturated with regard to this element, this explains why the intensity of this
»ne m the beginning of the exposure varies only slightly with the concentration.
Moreover it is to be remarked that the line has the same intensity in every part
of the arc, the line is a quot;longquot; line (Lockyer), it is not enhanced near the cathode.

Gradually the temperature is increasing, this means a higher ionization
and the possibility of emitting the radiations of wavelengths 4077.714 and
15.515 A. At the same time an unequal distribution of strontium in the arc
comes into being; of course the concentration increases everywhere, but prefe-
rentially near the cathode as a result of the attraction of the positive metal
ions to this place. Here the potential gradient is high enough to supply the energy
or the Sr II lines; these lines are much more sensitive in the region near the
cathode than in the gas column of the arc. Of course as a consequence of this
s ate of things, the concentration of neutral atoms is likewise greater in the
neighbourhood of the cathode than elsewhere and this fact explains the change
I character of the 4607.342line; it becomes likewise enhanced at the side of the
quot;egative electrode.

1 he lines appear stronger in the spectra of mixtures with sodium carbonate,
^is holds also for the quot;longquot; line as the darkness of the background is much
ss pronounced in these exposures, the CN-bands being almost entirely suppressed.

As has been mentioned on occasion of the discussion of the rubidium line

46nbsp;strontium, barium

at 4215.58 A, the place of the Sr line at 4215.515 coincides with the head of a
strong CN-band. The addition of sodium carbonate has not the effect of saving
this line from sinking in the heavy darkness of the negative at this place as it is
only being emitted after removal of the alkalies. With certainty the presence
of this line can only be estabhshed for concentrations down to 0.03 % SrO (300
g per ton), though it will of course be present even at much lower concentrations.

The carbons (and perhaps also the sodium carbonate) are not entirely free
from strontium. The quantity corresponds with a percentage of 0.0005 % in a
10 mg sample. This is not a serious disadvantage as amounts of several tenths
of a percent are not rare, of several hundredths of a percent quite normal and
quantities lower than 0.005 are seldom encountered.

Barium.

Among the more than sixty lines of the barium-spectrum between 3600
and 5000 A the following may be mentioned as being important for our purpose:

8R —
lOR —
lOR —

8 —

lOR — (Spectrograms 104 and 407)

Though the purity of the carbons with regard to Ba is fairly satisfactory,
the intensity of the strongest hne had to be valued with 3 in the case of no sub-
stance or only sodium carbonate being introduced into the arc. The intensity
scheme for the determination is:

The lines have decidedly a tendency to appear somewhat later than the
Sr-hnes; they remain visible to the last phase of the evaporation process. The
4579.66 line behaves comparable with the Sr-line 4607.342 and attains its full
intensity after 30 seconds. After 60—120 seconds the line gets the character of
a „shortquot; line but as it has a much smaller intensity this is less conspicuous than
in the case of strontmm. This line may safely be considered as an atom line,
hence: 4579.66 Ba I 8R.

It may be remarked that — obviously because the excitation potential of
barium is lower than that of strontium - the ion lines of barium are visible

barium, radium, zincnbsp;47

in the early phases as „longquot; lines and are then stronger than the corresponding
Sr lines, though as a matter of fact they get their full intensity only later as
enhanced lines.

In the spectogram of the pure barium compound (nr. 407) taken with and
without sodium carbonate the lines appear stronger in the exposures of the
undiluted substance and probably their intensity is proportional to the amount
of BaCOg introduced in the bore of the cathode. On the other hand, the spectro-
grams 450—459 furnish a striking example of the enrichment of the emitted
radiation when sodium carbonate is mixed with a base containing small amounts
of the element to be determined.

The quantity of BaO present in the carbon electrodes and the sodium car-
bonate corresponds with a percentage of somewhat less than 0.0003% in a 10 mg
sample. The amount of BaO found in our rock- and mineral samples was in
general higher than 0.001% so that I have not spent time and labour on trying
to purify the carbons in a more rigorous way.

Radium.

Gerlach and Riedl «) mention two lines of this element, most likely with
the reason of making their work more complete, as obviously it would be unwise
to determine this element according to other methods than those of radioactivity.
The lines are:

3814.44 Ra II 10 10 —
4682.20 Ra II 10 10 —

It may be remarked that other lines of the same intensity are reported in
Rayser's tables e.g. 4533.17 A.

Zinc.

. The sensitivity of the zinc determination with spectro-analytical methods
s comparatively low, not only in the visible spectrum, but also in the other
parts. 0.03% seems to be the limit and this is not improved by the addition
o sodium carbonate to the sample, on the contrary, this makes the zinc lines
isappear at higher concentrations so that zinc should be determined whenever
possible in the absence of substances which have this effcct and if this cannot
e done, the statements should be given with due reserve as the results likely

be too low.

llie usual tables arc in this case combined:

450 451 452 453 454
1 3.10-1 10-1 3.10-2 10-2

48nbsp;zinc, cadmium

Though apparently with high concentrations or when using electrodes of
the pure metal, the intensities of these three lines are identical, a pronounced
difference can be observed in the lower concentrations as shown in the above table.

The hnes appear in the very beginning of the process, during the first exposure
of 30 seconds and if 0.3% or more of ZnO is present they remain visible during the
second exposure also, otherwise theur duration is restricted to the first half minute.
In the presence of other easily volatile substances (alkaU and alkaline earth
compounds) the emission of the zinc radiation begins somewhat later besides
being repressed, as pointed out above.

Other zinc hnes mentioned in Kayser's tables are extremely weak in the
spectrum taken with pure zinc oxide:

4057.87 Zn 6 — —
4629.810 Zn I 8 — —

or do not appear at all even under these favourable conditions:

3739.96 Zn 4 1 —
4912.6 Zn II 10 10 —
4924.0 Zn II 10 10 —

All these hnes are rather strong in the spectrum of an arc between zinc
electrodes, with the exception of the Zn II lines for which I venture to doubt
the correctness of arc intensities as high as 10. These remarks are based on the
spectrograms 88 and 404 of our collection.

The atom Unes of zinc are scarcely enhanced near the cathode as during the
evaporation of the zinc the temperature remains too low to cause an effective
dissociation of the vapour.

Cadmium.

In spectrograms of a cadmium arc and of a carbon arc fed with cadmium
oxide the following lines have been.observed:

cadmium, mercurynbsp;^g

sample of cadmium was not of the utmost purity, nevertheless these lines seem
to belong to this element, at least I did not succeed in attributing them to an-

^^leTd n-nbsp;-d cadmium compounds are often contaminated

with lead, thallium and bismuth.

A pecuUar coincidence makes it impossible to use the lines at the short
end of the spectrum, at least when nickel may be expected to be present in the
sample. Now, the average amount of nickel in the earth's crust is of the order
of magnitude of 0.01%. for cadmium this seems to be 0.00001-0 0001«/ The
cadmium hnes are weaker than the nickel lines and consequently for the purpose
of our mvestigation this part of the spectrum is useless. This also applies to
the other end of the spectrum as cadmium can hardly be expected in such con-
centrations as are estimable by the method:

The other lines are too weak to be of any value.

suhnbsp;^^^nbsp;presence of alkalies and other easily volatile

herenbsp;decreases the sensitivity and the same precautions should be taken

The lines are visible during the first exposure of 30 seconds and also in the
typequot;nbsp;tban 10% of CdO is present. The lines are of the quot;longquot;

Mercury.

thenbsp;sufficient quantity of some pure mercury compound in the bore of

^ cathode — which may advantageously be taken much deeper than ordinarily
twelve hnes are registered on the spectrogram:

^^^^'•st the lines at 3906.44, 4108.07, 4216.7 and 4339. 21 A could not be detected.

mercury, boron, aluminium

It is not likely that the very small amounts of mercury commonly found in
1 Minpr.!. ran directly be determined with the spectrographic method

TnW^clS^

THIRD GROUP OF PERIODICAL SYSTEM.

TherTare no lines due to boron in the range covered by our spectrograph
(Spectrogram 37).

Aluminium.

In the arc spectrum of aluminium two quite conspicuous lines lie between
3600 and 5000 A:

3944.025 Al I lOR 8R —

3961.537 Al I lOR SR — (Spectrograms 67 and 424)

The lines at 3897.90 and 4150.138 A mentioned by Kayser could not be detected
even in the spectrum of an arc between metallic aluminium electrodes. In this
spectrum I found, however, a quite uncommon series of bands which do not
appear in any other of our numerous spectrograms, whether between metallic
electrodes or not. Consequently I am inclined to ascribe this spectrum to some
aluminium compound being formed in the arc, or else to another substance the
formation of which is perhaps promoted by the presence of aluminium. The heads
of the bands are at the high frequency side of the spectrum and will be found
approximately at: 4840, 4650, 4495 and 4330 A. In each band eight subordinate
bands can be distinguished which in their ^tum separate into some 27 lines.

aluminium, scandiumnbsp;51

In other metallic arcs only the common CN-bands have been observed if at all,
due to carbon from COg of the air or combined with the metal, e.g. in spectra
of purest silver from the Dutch Mint and of quot;physically purequot; platinum of
Messrs. H. Drijfhout amp; Zoon, Amsterdam.

Harking back to our proper subject, I have to remark that a spectrographic
evidence for the presence of aluminium in rocks wiU hardly ever be called for
as most rock types contain AI2O3 to an amount of 10 to 25%. Beyond these limits
lie at the high side only some rare alcaline rock types or magmatic rocks which
have assimilated sediments of the shale type, whilst low percentages are found
in many peridotites. Also, the carbon electrodes contain as much aluminium as
corresponds with 0.1 % in a common size sample, unless they are purified by bur-
ning an arc between them for some length of time; then the ends contain only
about 1% of the original amount of this contamination, judging from the inten-
sity of the lines.

In view of the reasons mentioned above I do not think it necessary to give a
table of intensities.

Scandium.

Very rich in lines is the spectrum of this element, more than fourty are
rather conspicuous when it is taken with the oxide and the number of weaker lines
is also great. It is difficult to make a choice which lines are to be mentioned:

52 and 435)

scandium

The mutual position of these and of a number of weaker Unes is given in

^'quot;quot;Though having the same intensity when high eoneentrations of scandium
inougn lid. 5nbsp;^^ different concentrations. The

compounds are ^^^^nbsp;only restricted by the possibility of coin-

elements, further there is but Uttle choice between
cidences ^itn imnbsp;spectrum apparatus is large enough to se-

Xiron ine interferes, it remains possible to make use of the sensitive scandium
line W an analogous speculation as has been mentioned when deahng with ni-
KH im In that case it is desirable to confirm the estimation by companng
Xr lines of reference negative and spectrum of sample under discussion. The
dispersion of our spectrograph is amply sufficient to dispense with this shghtly

obiectionable procedure.nbsp;• r

The foUowing table of intensities has been computed from a senes of spec-
trograms numbered 480—489.

Spectrogram . . . .
Percentage of ScjOa .

480 481 482 483 484 485 486 487 488 489
10 1 3.10-^ 10-1 3.10-2 10-2 3,10-3 10-3 3J0-* 10-*

The Unes 3630.75 and 3642.81 are out of the question as already pointed out
by Goldschmidt and Peters in their paper on the geochemistry of Scandium
These Unes are very sensitive, but coincide respectively with 3630.73 Ca I 6 and

with 3642.680 Til lOR.

Owing to the low volatility of scandium compounds, the Unes are not shown
in the first exposure and gradually increase in intensity during the foUowing phases
of the evaporation. Ninety to hundred seconds after striking the arc they have
reached their maximum intensity remaining there for the rest of the work if this
is not unduly prolonged. An exception is to be made for pure Sc^Og mixed with
50% of sodium carbonate by weight; here the emitted radiation is strongest

♦) Used by Goldschmidt and Peters in their communication on the geochemistry
of scandium, together with a number of lines in the more remote ultraviolet.

scandium, yttrium

directly after ignition of the arc decreasing slowly after the second exposure of
thirty seconds.

It is highly advantageous to mix the sample with a sufficient quantity of
sodium carbonate, e.g. the spectrum of a sample containing 0.03% ScjOa without
addition of sodium carbonate has a somewhat lower brightness than the spectrum
of a sample containing 0.01 % ScjOg with sodium carbonate. Apart from this effect,
the nature of the base is essentially without influence on the intensity of the lines,
as stated already by Goldschmidt and Peters. This is to be expected as the scan-
dium lines only appear after the volatilization of a large part of the base substance.
The cathode layer effect is very pronounced.

The carbons and the sodium carbonate used for the dilution of the samples
proved to be entirely free from scandium.

Yttrium.

No attempt is made to enumerate the countless yttrium lines appearing on
the spectrograms of our collection. In the table below and in the figure 8 are listed
only the lines of importance for spectrum analysis, which are eight in number for

After excluding two lines which arc likely to yield difficulties owing to
coincidences with lines of other elements (4077.714 Sr II lOR is often rather
strong and then masks the surroundings, 4374.51 Sc II 10 and 4375.00 Nd 10)
the following table of intensities results:

yttrium, lanthanum

Though generaUy speaking there is much resemblance between the be-
haviour of scandium and yttrium during the evaporation process m the arc
there are some minor divergences which may be mentioned here, whilst further,

reference is made to the foregoing section.

The Hnes of yttrium are in general much weaker than those of scandium as

a result of the emission of a great number of wavelengths. They begin to appear
somewhat later than the scandium lines (this may be only an apparent difference
caused by their lower intensity) and likewise reach later heir full brightness (t^
sneaks for a real later appearance). The character of quot;enhanced lines is still
stronger than in the case of scandium, also when pure yttrium oxide is being arced.

The influence of base substance on the intensity of the lines was found
to be negligible by Goldschmidt and Peters (p. 261) for as widely diverging
substances as SiO^ and CaO. I can confirm this statement, the addition of sodium
carbonate, which ordinarily strongly changes the intensities, has a less marked
effect on the emission of the yttrium lines. This may be explained in the same way
as has been done for scandium and evidently, the later the lines appear, the smaller

will be the action of the flux.

The electrodes used for the investigation and the sodium carbonate were

free from yttrium.
Lanthanum.

Of the numerous lanthanum lines the following are mentioned as being
especially useful for the purpose of spectrum analysis: