Professional Documents
Culture Documents
I requirements of modern
industrial operations, a
laboratory must be in a posi-
estimation of small amounts of i m
constituents in materials is PO
advantages are (1) speed, (2)
a l l o y s whose physical and
chemical properties, such as
hardness, ductility, and re-
sistance to corrosion, are pro-
tion to turn out analytical re- certainty in the identification of foundly modified by appar-
sults in the shortest possible determined. Illustrative exam ently insignificant traces of
time, consistent with reason- kinds of work for which they may be used. i m p u r i t i e s , In many cases
able accuracy. Too often the The general principles of quantitative methods a t h e causes of variations in
result of a p e r f e c t l y good discussed, certain sources of error are noted and mea these properties have not been
analysis is of value only as an suggested for overcoming them. recognized because of the
explanation of defects and I Methods which have been developed in the labor p r a c t i c a l impossibility of
difficulties, rather than as a tory of The New Jersey Zinc Company are describe making the necessary analy-
guide in the production or These methods are used in carrying out several hundr ses. In such cases the spec-
selection of the material in- analyses each month, many of them being routine con- \ trograph provides the means
volved. This is particularly trol analyses. both for tracing these causes,
true with analvses for those and, when thev are known,
minute amounts of impurities, or of necessary constituents, for controlling the variations by proper selectibn of raw ma-
which often modify to a remarkable extent the chemical or terials. In this way it has been possible to show that signifi-
physical properties of materials. cant and reproducible effects are caused by the presence in a
This being the case, it is surprising that so little attention metal of less than 0.001 per cent of some impurity whose
has been paid t o the spectrograph as a means for carrying presence had not been suspected until it was revealed, and its
out quantitative analyses. I t has long been recognized as amount estimated, by means of the spectrograph.
an instrument for qualitative work, but even in this field its For such sinall amounts as these, ordinary chemical meth-
usefulness has not met with the general recognition which it ods of analysis are often unsatisfactory. Large samples are
deserves, particularly in industrial laboratories. Its remark- required, necessitating the handling of large volumes of solu-
able sensitiveness and the certainty and speed with which tions. Separations are incomplete, involving numerous re-
many of the elements niay be detected place it in a class by solutions and re-precipitations before it can be assumed that a
itself, far superior, in many respects, to the ordinary methods precipitate is free from contamination- an assumption which
all too often is not justified by the facts. With the large vol-
umes of solutions used, the very slight solubility of nominally
insoluble precipitates may cause significant errors. Finally
Type of Work for Which Adapted the presence of unsuspected constituents, with analytical
properties similar to those of an element which is being de-
Several types of analytical work fall particularly within its termined, may lead to totally erroneous results. Technical
scope: methods of analysis ordinarily provide for the removal of only
1-Quantitative estimations of traces of impurities and
the most common interfering elements. As a result cases
minor constituents, when the amounts present are too small have been encountered in which a final precipitate, weighed
for satisfactory chemical determination. and calculated as a compound of one element, has been found,
2-Rapid estimation of elements, present in small amounts, by spectral analysis, to contain none of that element whatever,
which could be determined chemically, but only by complicated but to be really a compound of another element whose pres-
and slow methods.
3-Determination of the approximate composition of materials ence was not anticipated when the method used was devised.
when the amounts available for analytical samples are too small RAPIDESTIBIATIONS-In the second class occur such de-
for chemical analysis. terminations as that of cadmium in zinc (especially in high-
grade zinc, containing less than 0.01 per cent of cadmium).
I DETERMIYATIOK OF IMPURITIES, ETc.-cnder the first To insure accurate results by chemical methods, a long and
' heading may be included the analysis of many metals and time-consuming series of separations is necessary and the
);. 1 Received June 7, 1928. analysis usually requires several days. I n contrast with
--.
2 ANALYTICAL EDITION Vol. 1, No. 1
this, the spectrographic method requires but a small sample, tensity will have become weak, and strong lines will have
it involves no chemical separations whatever, and yields just decreased in intensity. Still greater dilutions result in the
as accurate results in an hours time. Another typical ex- disappearance of other lines and the further weakening of
ample in this laboratory was the emergency analysis of five those that remain, until, at the lowest concentration a t which
samples of zinc-base alloys for tin, magnesium, cadmium, and the element can be detected at all, only one or two faint lines
lead-all within 3 hours. At the same time it was definitely remain. These most persistent lines of all are called, in the
established that all other elements, which might have had a terminology of de Gramont, the raies ultimes of that element.
bearing on the particular problem then under investigation, If the conditions of excitation and of observation are kept
were absent. At least 2 Qr 3 days would have been required perfectly constant, and if the samples used are of similar
had these been determined by ordinary methods. character and size, these variations will always take place in
Similar savings in time are possible in the determination the same way. I n other words, the same lines with the same
of many other elements. I n addition to those already ...,-
intensities alwapbe~foundin the spectrugof T i2
-
mentioned, lithium, sodium, copper, silver, beryllium, alumi-
num, thallium, antimony, bismuth, manganese, and nickel
-
in-*question, w F i t isupresent a t the same concentration. A
quantifatiz a n G i s can therefore K?made by com-
have all been estimated-mostly in alloys and other zinc- paring the number and intensities of the lines of the element
bearing materials. Any element which yields a spectrum to be determined, in the spectrum of the sample with those
containing lines which permit its identification in the particu- of a series of reference standards.
lar material involved can be determined in an hour or two.
Most of them require as many days for estimation with equal Elimination of Variables
or less accuracy by chemical methods, when present in the The most important factor in determining the precision
small amounts for which the spectrograph is particularly which may be attained in such an analysis is, of course, the
adapted. constancy with which all the variables involved may be con-
ANALYSIS OF MINUTE SAiwLEs-The third class, the analy- trolled. Some may be kept within very close limits; others
sis of minute amounts of materials to obtain a general are subject to slight unavoidable variations. The most im-
idea of their composition, includes a large field of usefulness, portant point in devising the technic to be followed is, there-
and one in which the spectrograph is nearly indispensable. fore, to minimize, as far as possible, the range of these varia-
It is often necessary to know the principal constituents and tions, or else to conduct the analysis in such a way that the
their approximate ratios when only extremely small samples same variations occur in the comparison standards and in
are available. Sometimes it may be a stain or a film of the spectrum of the sample to be analyzed. The relatively
dust or discoloration on the surface of material. Again it low degree of precision attained with spectrographic methods
may be a trace of precipitate, too small for chemical identifi- by some authors may be ascribed in large measure to a failure
cation. At other times it is necessary to make analyses with- to realize that such compensation for unavoidable variations
out defacing rare or valuable specimens. is possible.
Many other instances could be given, but those described Two methods have been proposed for this purpose. Ger-
are sufficient to indicate some of the ways in which an instru- lachlo and Schweitzer,ll instead of comparing directly the
ment of this kind can furnish information which could only intensities of the spectrum lines of the elements to be deter-
be obtained with difficulty, if a t all, by other means. mined with those of the same lines in standard spectra, use,
Principles of Quantitative Spectral Analysis as reference standards, lines in the same spectrum, due to
the principal constituent of the sample. This constituent
The foundations of quantitative spectral analysis were laid is assumed to be constant in amount, or to vary only as a
by Hartley.2 His work with that of his successors, Pollok result of the variation of the element to be determined. Thus
and L e ~ n a r djn
, ~ Dublin, and of de G r a m ~ n t ,in~ ,Paris,
~ has in analyzing for small amounts of tin in lead, the tin lines are
been very fully reviewed by several recent writerse6to compared with certain selected lines of the lead spectrum.
Briefly, the principle involved in the methods used by all From previous experiments it is determined, once for all, that
these investigators is that of the gradual weakening and under the conditions prevailing a certain line in the tin spec-
simplification of the spectra of all the elements, when the trum is equal in intensity to one of the neighboring lead lines
amounts present in the light source, whether flame, arc, or at some one concentration, and to certain other lead lines a t
spark, are decreased. Thus if a pure element or a very, con- other concentrations. Thus, under the conditions described
centrated compound or mixture is used, the spectrum is made by these authps, at 10 per cent tin the tin line of wave
up of a considerable number of lines of different intensities. length 02572 A. is equal in intensity to the lead line a t
If, now, a second spectrum is observed, using a dilute mixture A2629 A., while at 0.6 per cent it is reducedin intensity and is
of that element with other materials, its spectrum lines will then equal t o the weaker lead line a t A2657 A. At the inter-
be found to have decreased, both in intensity and in number. mediate cgncentration, 5 per cent tin, another tin line, that
I n general, those lines that were weakest in the first spectrum a t A2422 A., has become equal to the lead line at A2412 8.,
will have faded out completely. Lines of moderate in- while with 0.04 per cent tin the tin line at 42707 A. has the
* Hartley, J . Chem. SOC.33, 210 (1882), Phzl. Trans. ROY. Soc. same intensity as the lead line of A2657 A. I n this way
(London), 1884, pt. 1, p. 50; pt. 2, p. 327; Proc. R o y . Soc. (London), 69A, after a table has been prepared indicating the concentra-
283 (1902). tions at which selected pairs of lines, one for each element, are
8 Pollok and Leonard, Sci Proc. Roy. Dublin Soc., 11 (N. S.),184, 217,
229 (1907); 257,270 (1908); 16 (N. S ), 273 (1918).
of equal intensity, a simple inspection of the spectrum of the
4 De Gramont, numerous papers in the Comgf. rend., 144 to 176 (1907 unknown will permit a fairly close estimate of the concentra-
to 1922, incl ). tion, This method assumes that, inasmuch as the lines of
6 de Boisbaudran and de Grarnont, Analyse Spectrale Appliquee the element to be determined and the reference lines are
aux Recherche de Chimie MinCrale, Librairie Scientifique J. Hermann,
Paris, 1923.
present in the same spectrum and photographed at the same
6 Twyman, Wave-Length Tables for Spectrum Analysis, A. Hilger, time, any variations in the conditions of excitation and pho-
Ltd., London, 1923. tography will affect all lines equally and thus eliminate any
7 Twyman and Smith, Am. Inst. Min. Met. Eng., Tech. Pub. 79.
errors due to these variations.
8 Meggers, Riess, and Stirnson, Bur. Standards, Sei. Paper 444.
9 Porlezza and Donati, Ann. chim. applicala, 16, 519, 622 (1926); 10 Gerlach, Z. anotg. allgem. Chem , 142, 383 (1925).
17, 3, 14 (1927). 11 Schweitzer, Ibzd , 164, 127; 166, 364 (1927).
January 15, 1929 I N D UXTRIAL A N D ENGINEERING CHEMISTRY 3
and P b
and P b
and P b
and P b
vfg aind Pb
5
0
m
N
Figure 1-Portion
z N
U c
N
of Preliminary Standard Plate for Estimating Magnesium a n d Lead in Metallic Zinc
This method offers distinct advantages in the analysis of Since no two plates are strictly comparable, because of
simple mixtures of elements whose spectra consist of many variations in the emulsion as well as in the composition, age,
lines with a. wide variety of intensities. On the other hand, and temperature of the developer, this comparison is not
where the principal constituent of the material to be analyzed relied on for the final determination except in those cases
yields a spectrum with relatively few lines, and these separated where a rough estimation is sufficient. The final plate is
by considerable wave length intervals, it might be difficult to usually taken with five spectra, close together, but not quite
find enough lines to serve as standards in the near neighbor- in contact with each other. (Figures 2 and 3) The first
hood of suitable lines of the elements which were to be de- one is that of a standard with slightly higher concentration
termined. Furthermore, the presence of three or more vari- of the element to be determined than was indicated by the
able constituents would be very likely to complicate the prob- preliminary test. Next comes the spectrum of the sample,
lem to such an extent that the accuracy of quantitative esti- followed by that of a second standard, of about the estimated
mations in complex materials would be subject to considerable concentration. A second spectrum of the sample comes next,
doubt. and finally a standard of slightly less than the expected value.
The second method, which has been used successfully in In this way the intensities of the lines in the sample spectra
this laboratory for several years, is to compare the intensities are quite sure to fall somewhere between those of the high
of the lines of the elements to be determined in the spectrum and low standards. It is then a simple matter to estimate the
of the unknown sample with those of the same lines in a true concentration in the sample, free from errors due to
graded series of standards of similar composition and photo- variations in plates and development.
graphed under similar conditions. A preliminary rough The usual practice is to use a series of standards whose con-
estimate is first made, in which a spectrum of the sample, on centrations with respect to the element in question vary by
one plate, is compared with a series of standards, previously multiples of 2 (0.1, -0.05, -0.025 per cent, etc.). These
photographed on another. (Figure 1) On this preliminary show differences between successive standards which permit
standard plate the successive spectra are spaced so as to leave the visual estimation of several intermediate degrees of in-
bettween them clear strips, slightly wider than the individual tensity. Nothing is gained by using closer intervals between
spectra, so that the sample spectrum may be placed between standards. It is helpful to use a rather wide slit in taking
adjacent standards and a better comparison made than is these spectra, as this gives an appreciable area to the lines
possible when the spectra are a t some distance from each and makes it much easier to judge intensities than when the
other. The plate with the sample spectrum is placed over usual narrow slit used in qualitative work is employed. Care
the standard plate, and moved up or down until an approxi- must be taken not to use so wide a slit that the lines to be
mate match with the standards is found, using a medium- observed overlap other lines in the spectra.
power magnifying glass for observing the lines. In this way Attempts have been made to use a microphotometer for
the concentration is approximately determined by compari- measuring line intensities, but no real advantage is gained.
son of the lines of the constituent to be determined with those This is probably because, in spite of all the precautions that
in the two adjacent standards. are taken to insure the constancy of conditions, there are
4 ANALYTICAL EDlTI0,V Vol. 1, No. 1
n
Mg -lA41:
IIIII
d-O.?% Mg
d-0.1% Mg
d-0.05% Mg
still uncontrolled fluctuations, especially in the light source, w. Those of next higher intensity a t 2803 and 2796 1.are
which produce variations of about the same relative magni- far too strong to serve for comparison.
tude as the errors in the visual estimation of intensities.
Duplicate determinations by the method described ordinarily Table I-Comparison of Chemical a n d Spectrographic Analyses of
Zinc a n d Zinc-Base Alloys
vary by less than 10 per cent of the amount being estimated ELE- ELE-
from the mean, or from the results of careful chemical analyses MENT CHEX- SPECTRO- MENT CHEX- SPECTRO-
AXALY-DETER- KAI. GRAPHIC ANALY-DETER- ICAL GRAPHIC
and this is probably as good as can be expected from ordinary SIS MINED ANALYSISANALYSIS SIS MINED ANALYSIS ANALYSIS
technical methods of chemical analysis on constituents present P e r cen2 P e r cent P e r cent P e r cent
to the extent of less than 0.5 per cent, which is the range for 1 Pb 0 052 0,050 11 Cd 0.0078 0.0070
2 Pb 0 054 0,060 12 Sn 0.0017 0,0006
which spectrographic methods are best fitted. 3 Pb 0 039 0,033 13 Sn 0.0005 0.0006
4 Pb 0 038 0.040 14 Sn 0.25 0.26
5 Cd 0 023 0.022 15 Mn 0.0062 0,0043
Comparison of Chemical and Spectrographic Analyses 6 Cd 0 020 0.025 16 Mn 0.0083 0,0050
7 Cd 0 021 0,026 17 Mn 0.0043 0.0033
8 Cd 0 010 0.007 18 Mg 0.13 0.16
Table I gives some idea of the agreement that may be 9 Cd 0 050 0.055 19 Cd 0.005 0.002
expected between chemical and spectrographic analyses. The 10 Cd 0 0017 0.0013 0.0026
repeat
figures were taken at random from records of routine work
with samples on which both methods were used. In reading Table 11-Repeated Analyses for C a d m i u m on S a m e Sample
them it should be kept in mind that. except for the higher PLATE OBSERVER1 OBSERVER2
leads and cadmiums, the spectrographic method was adopted P e r cent P e r cent
0.035 0.03
because chemical methods had proved unsatisfactory. For 0.04 0.03
0.04 0.04
this reason differences cannot be considered as indicating 0.04 0.036
which one is in error. A better basis for judgment is to be 0.04 0.04
0.04 0.04
found in the consistency with which results can be duplicated Average 0.039 0.036
on the same sample. This is shown in Table 11,which gives,
for a single sample, two series of estimates made at different At a concentration around 0.01 per cent (Figure 3) the
times and on different portions of the sample by two independ- group of five lines hasobecome too faint to serve, while the
ent observers. Observer 1 prepared the samples and photo- pair a t 2803 and 2796 A. is still too strong when the same ex-
graphed the spectra, and knew what results had previously posure is given to the plate. As there are no suitable lines
been obtained. Observer 2 knew nothing of the identity of intermediate intensity within this spectral range, recourse
of the samples when grading the plates. They came to is had to a rotating sector wheel with but a 5 per cent opening
him simply as part of the daily routine, the only infor- between the arc and the slit. This gives about the right
mation given him being the element to be estimated and line intensity for the pair, with the same total time of ex-
the concentrations of the standards used in preparing the posure, which is necessary t o insure the complete volatiliza-
plates. tion of the sample. With this exposure the raie ultime of
Choice of Lines to Be Compared magnesium at wave length 2852 8. is also weak enough to
serve.
The choice of the lines to be used when making comparisons If the attempt is made to base the estimates on too strong
requires some consideration. The impression might be lines, serious errors may be made. Many strong lines exhibit
gained from a casual reading of the literature that certain &versa1 and this sometimes appears as a weakening of the
lines have peculiar properties which make them more signifi- whole line, when the absorption is not strong enough to show
cant than others, or that they show more marked changes in complete reversal, with a clear centered line on the plate. It
intensity for a given change in concentration. Except for is possible, therefore, to find lines in the spectrum of an ele-
those extremely diffuse lines which are so ill-defined a t low ment, present a t high concentration,which appear weaker than
intensities as to make them difficult to observe, this is not in the spectrum of the same element a t considerably lower
the case. All lines go through a sensitive range in that stage concentration. It may be that this effect is the real basis
of dilution just before they completely fade out. In this for the statement, originally made by Hartley2 and repeated
range the line is of some shade of gray sharply defined and by nearly all subsequent writers, that the strongest lines in
free from halation or "wings." Different lines, or groups of the full spectrum of an element are not always the ones which
lines, will therefore serve for comparisons a t different concen- persist to the lowest concentrations. This has been vigor-
trations. The raies ultirnes can only be used a t the very low- ously denied by Gerlach'o who maintains that the strongest
est. I n Figure 2, in which the magnesium is of the order of lines are always the most persistent. A safe rule, therefore,
0.1 per cent, the estimation is best made on the basis of the is to base all estimates on the weakest lines that are clearly
group of five weak lines between wave lengths 2783 and 2777 defined.
-
January 1.5, 1929 I N D GSTRIRI, AND EiliGIiVEERI,VG CHEMISTRY 5
M g
Mg
Figure 3-Final Plate for Estimating Magnesium i n Zinc-Base Alloy (Mg 0.009 Per Cent)
The question is frequently asked-how high concentrations Most of the workers in this field have made use of the high-
can be determined spectrographically? The answer is that tension spark, exclusively, as the source of light.%3, 4, *, 10, 11
there is no limit, except that the errors of estimation will be Lewis'* states, however, that the tendency in technological
proportionately large with high percentages. The relative circles is to use the arc, as it is quicker and more convenient
precision is the same, whether the concentration is high or than the spark and gives a fair degree of accuracy. Porlezza
low. This means that spectrographic methods cannot com- and Donatig are also using the arc in their extensive study of
qete in precision with chemical in the determination of those quantitative spectrography. Twyman and Smith7 state
elements which form the main bulk of a material. that, in their opinion, the arc is the most generally useful form
It is a t the other end of the scale, in the range of extremely of excitation. The author prefers and uses the arc for most
small fractions of 1 per cent, that the spectrograph finds its purposes, although some elements, such as selenium and tel-
greatest usefulness. Some elements, such as magnesium, lurium in very small amounts and most of the non-metals and
bismuth, beryllium, or copper, can be determined with con- gases, can best be detected by means of the spark or vacuum
centrations as low as a few parts in ten million. Others, tubes. No attempts have been reported, as far as the author
however, are much less sensitive. Arsenic can just be de- is aware, t o use the spectrograph for quantitative gas analysis,
tected by the method described a t one part in ten thousand. although it appears to have some possibilities. It is wise,
In the case of uranium the limit is higher still. therefore, when equipping a laboratory, to make provision
The amount which can be detected also depends somewhat for both methods of excitation. The main advantages of
on the other elements present. A dilute solution of a lead the arc are its greater sensitiveness and greater speed, both
salt in water, with no other metals present, yields a spectrum being due to the extremely high temperature of the electrodes
in which the lead lines are much weaker and fade out, on di- and the consequently greater density and concentration of
lution, a t a higher concentration than one containing the the vapors evolved.
same amount of lead and a considerable amount of zinc.
Electrodes
Serious errors were made during one stage of the development
of this method by failing to include aluminum in the standards Carbon electrodes are preferable, in most cases, to those
used in analyzing, for lead, a zinc-base die-casting alloy con- composed of the material to be examined, even when that is
taining 4 per cent of aluminum. The presence of the alumi- a metal or other conducting substance. This is because the
num in the alloy intensified the lead lines so that they were carbon arc is by far the hottest and will therefore reveal
nearly twice as strong as those of the equivalent standards. smaller traces of impurities than will the arc between metal
When the corresponding amount of aluminum was included electrodes.
in the standards, their lines also were intensified and correct One American manufacturer is preparing specially re-
rcsults were obtained. This simply emphasizes the necessity graphitized rods for this purpose, which are remarkably free
for using standards of very nearly the same composition as from impurities, whose spectrum lines might interfere in
the materials to be analyzed. the ordinary run of analysis. Traces of titanium, vanadium,
Limitation of Standards
silicon, magnesium, calcium, and boron are usually present,
but not in sufficient amount to be troublesome, except when
The objection is sometimes raised that this method involves the very smallest traces of these elements are to be determined.
the preparation of a very large number of standards. While A British optical firm furnishes a grade of carbon which is
this is true, it is also possible, by limiting the field of activity free from titanium and vanadium and is therefore useful
to those analyses for which no satisfactory rapid chemical when small amounts of those elements are to be detected or
methods are available, to keep the number of standards estimated. On the other hand, this carbon contains as
within reasonable bounds. Many standards can be prepared much or more of the other impurities in the list given than
containing more than one variable constituent, so that one does the graphite and, in addition, appreciable amounts of
analysis will serve for the determination of several elements iron, copper, and manganese from which the graphite is free.
at the same time. Most standards can be kept indefinitely, The ordinary commercial arc-light carbons are of very limited
and as only minute amounts are used for each analysis, once usefulness, as all of these, together with other impurities, are
prepared, they last a long time. Most important of all, the present in larger and more variable amounts than in either of
saving in time that results from the use of spectrographic the other grades.
methods is so great that it more than justifies the trouble In one respect the graphite is inferior to both the British
involved in the preparation and care of the necessary stand- and the common arc-light carbons. I t will not hold as steady
ards. 12 Lewis, Cantor Lectures, Royal Society of Arts (London), April, 1921.
6 A N A L Y T I C A L EDITION Vol. 1, No. 1.
an arc. With the graphite the arc has a tendency to wander a method such as this, which will detect segregation of im-
from point to point on the surface of the electrode. This, purities has very much to commend it is perfectly true if
of course, is highly undesirable, as it tends to give variable that is the object of the analysis, but is certainly not valid
illumination on the slit and thus to yield erratic results. Two in the case of the usual quantitative analysis, whose object is
methods have been used to overcome this difficulty. Rapid to ascertain the average composition of a larger quantity of
rotation of the upper electrode helps considerably, but the material. Furthermore, the use of wedge-shaped electrodes,
best results are obtained when a stream of air is drawn in at relying on the wandering of the discharge along the edge to
the tip of the upper electrode through an axial hole. The insure a true average, is by no means certain to achieve that
electrode fits tightly in a special holder which is connected result. Cast metals frequently contain single crystals an
with a vacuum line, through a bottle arranged to insure con- inch or more in diameter; therefore, electrodes prepared from
stant suction. This device helps in two ways to improve them could not be relied on to yield an average spectrum.
the constancy and reproducibility of the light source. It
not only reduces the tendency of the arc to wander, but i t Preparation of Sample
also draws into the arc the vapors liberated from the sample
material on the lower electrode, and makes it more certain Several years experience in the analysis of this kind of
that no vapor can escape without recording its presence by material has served to emphasize the necessity of choosing
its radiation. a sampling method that will insure that every portion of
The principal objection to the use of any form of carbon for material whose spectrum is registered on the plate shall be
electrodes is the presence, in its spectrum, of the very intense as nearly as possible truly representative of the main bulk of
cyanogen and other bands in the visible and longer wave- the material.
length part of the ultra-violet regions. These make it very The amount of sample necessary for a single determination
difficult to detect weak lines of other elements in the same is ordinarily from 20 to 50 mg. This is much too small an
regions or to compare intensities when they can be detected. amount to obtain by cutting down the solid material, even
Fortunately, however, there are very few elements for whose in the form of an extremely fine powder. The safest method
detection and estimation lines in this range have to be used. is to use solutions. If these are properly preparedand mixed,
The ultra-violet, from about 2350 to 3400 A., contains lines there can be no question of uniformity.
of most of the elements, of sufficiently varied intensities to Hartley, Pollok and Leonard, de Gramont, and others em-
serve for estimating small amounts. Moreover, the higher ployed solutions in much of their work. I n their method one
dispersion and resolving power in this range, with a prism in- of the spark electrodes (a platinum or gold wire) is immersed
strument, make it by far the more generally useful. in the solution, the other being placed directly above it. A
The spectrum emitted by the carbon arc, between these bit of capillary tubing, of glass or silica, surrounds the lower
limits, when photographed with an instrument of high dis- electrode and projects slightly above its tip and above the
persion and resolving power, is made up of an almost con- level of the solution in the containing vessel. Capillary at-
tinuous band of faint, closely spaced lines. Fortunately, traction then raises the solution in the tube above the sur-
these are of very nearly equal intensity throughout, so that rounding level, so that the tip of the electrode is submerged.
even when coincidences do occur with lines of other elements, I n this way the spark passes between the upper electrode and
there is no serious difficulty in detecting the latter by the in- the surface of the solution in the capillary and is high enough
creased intensity of the lines, in comparison with the neigh- above the surrounding solution so that the path of the light,
boring electrode background, and with an adjacent blank from the spark to the condensing lens, is not obstructed by
spectrum from the bare electrodes. the solution. A modification of this, using graphite rods in
In exceptional cases it may be desirable to use electrodes of place of platinum wires for electrodes, has also been sug-
materials other than carbon. Copper is sometimes used, as gested.
its melting point is high enough to withstand the temperature A somewhat similar arrangement has been used, a t times,
developed by the arc without melting down. It is sufficiently with the arc, but is not nearly so satisfactory as the method
free from impurities, even in the ordinary electrolytic grade, which has been developed in this laboratory for a great variety
to serve for most purposes, and it does not oxidize so rapidly of routine determinations. This method is to dissolve the
as some of the other metals that might be used. Also, its sample (which should be of such size, and obtained in such a
spectrum is composed of fewer lines and the chances are less may that it conforms to the usual laws of sampling) in ap-
that lines will be found, both from the electrodes and from the propriate acids or other solvents and to adjust its volume in
sample, so close together as to interfere. some simple ratio to the weight of sample used. When the
Many authorities use, as electrodes, rods of the material concentration of the constituent to be determined is small,
to be analyzed, when that is a metal or other easily conducting the volume should be small, too, so that the solution may be
material, and cause the spark or arc t o play between them. as concentrated as possible.
This would be the ideal way were it not for two objections Procedure for Analysis
which are more or less serious, depending on the material
involved. Many of the metals have such low melting points A measured amount, usually 0.1 cc. of this solution, is
that they can only be used as arc electrodes if the current is dropped from a capillary pipet into a hole in the upper end of
very weak. This means low light intensity and long photo- a graphite electrode. A convenient size for the electrodes is
graphic exposures. This objection is not so serious as the 8 mm. in diameter and 50 mm. long. Prior to adding the
other, which is that many metals are far from being homoge- solution, the electrodes are prepared by burning them for a
neous. Zinc, for example, on cooling from the melted condition minute or two in a 6- or 8-ampere arc. This develops a cer-
is very liable to form large crystals, with most of the impurities tain amount of porosity and at the same time helps to expel
segregated in the intercrystalline boundaries. It would thus traces of impurities from the graphite. The electrodes are
be very improbable that the particular points between which then cooled before adding the solution. After the solution
the discharge played would be accurately representative of has been added, the electrode is dried in an oven. The im-
the average composition of the bulk of the material. pregnated graphite is used as the lower electrode in the arc,
The contention in a recent paper13that on the other hand, the upper one being also of graphite, arranged as previously
18 Twyman and Smith, Zoc. cit., p. 11 described, to permit air to be drawn through it.
January 15, 1929 INDUSTRIAL A N D ENGINEERING CHEMISTRY 7
A quartz condensing lens serves to concentrate the light can resolve closely coincident lines of two elements, which
from the arc on the slit and to exclude that from the incan- must be separated to permit the certain identification of one
descent carbons, whose strong continuous spectrum would be constituent and its estimation, free from the possibility of
objectionable, Suitable adjustments are provided, which error due to an overlapping line of another. I n general,
permit the arc to be moved both horizontally and vertically therefore, any laboratory which has a large amount of work
so as to keep it properly centered during the exposure. Con- adapted to the spectrographic method will find that the
stant arc length is insured by keeping the tips of the electrode additional cost of the larger instruments is fully justified by
images always a t the same points above and below the aper- the greater certainty of the results which may be obtained
ture in the slit diaphragm. With a reasonably constant line and by the wider range of possible applications.
voltage in the power supply, this control of the arc length is Three-minute exposures are given with most materials,
all that is necessary to maintain constant current and voltage using an arc carrying 10 amperes with a 60-volt drop across
drop in the arc. the arc terminals. This length of time is sufficient to insure
I n order that the operator's attention may be concentrated the complete evaporation of the sample, or at least the expul-
on keeping the arc properly centered, an automatic timing sion of all of the constituents to be determined, It is quite
device has been installed. This consists of an electric clock so necessary that none be left in the electrode, otherwise erratic
connected through a system of relays that it operates only results will be obtained.
while the arc is burning, and extinguishes the arc as soon as Slow photographic plates give the best results. Their
a predetermined exposure time has elapsed. This is particu- fine grain makes it easy to compare line intensities, and their
larly valuable in photographing the spectra of very volatile freedom from fog is an important consideration, particularly
materials, which often yield almost explosive arcs, difficult when comparing extremely weak lines.
to keep burning continuously. If it were necessary for the
operator to keep track of the intermittent exposures that Conclusion
result under such conditions, he could not give proper at-
tention to the centering of the arc, which is one of the most im- No hard and fast rules can be given for methods applicable
portant factors in insuring that all spectra shall be accurately to the determination of all elements in all kinds of materials.
comparable. Some of the general methods have been sketched, which
Using a large Littrow-type quartz spectrograph, with an have been found useful in a laboratory devoted mainly to
objective of 70 mm. aperture ando170 cm. focal length, the the study of problems connected with the production and use
spectrum from A2350 to A3400 A. is photographed on a of zinc and zinc-bearing materials. Each type of analysis
plate 10 inches (25.4 cm.) long. This is the spectral range that requires individual study to find the most suitable details
has been found the most generally useful for much of the work of technic. The main purpose of this paper is to call to the
that is encountered in the average metallurgical laboratory. attention of the chemists of this country the apparently little-
Other ranges may be utilized, if desired, by a simple change appreciated fact that in the spectrograph they have a power-
in the adjustment of the instrument, but experience has shown ful tool, capable of yielding information of the greatest value,
that a t least 95 per cent of the work can be done within these with a speed and certainty that cannot be matched by any
limits. Persistent lines of most of the elements are found other means. A few laboratories have installed the necessary
there, and the dispersion is great enough to insure accurate equipment and are finding new uses for it almost daily. In
identification of lines. The most complete and accurate lists that of The New Jersey Zinc Company several hundred
of lines are available for this region, and ordinary plates are spectrographic analyses are made each month, mostly quanti-
sensitive to these wave lengths without special treatment, tative estimations, and many of them are daily routine and
such as bathing in sensitizing dyes or fluorescent oils. control analyses.
Smaller instruments can be used for a great deal of work.
Several makers put out quartz spectrographs in which the Acknowledgment
full length of the photographable spectrum is included within
the length of a single 10-inch (25.4-cm.) plate, and excellent Grateful acknowledgment is made to George W. Standen,
results can be obtained with them on many materials. Oc- whose painstaking care and attention to details have been
casions do arise, however, when only the larger instrument important factors in the progress of this work.
MELLONI N S T I T U T E O F I N D U S T R I A L R E S E A R C H , U N I V E R S I T Y OF P I T T S B U R G H , P I T T S B U R G H , PA.
tions exceeding the range of those between duplicate de- not exhausted by the 0.5 cc. of reagent. It is clear, therefore,
terminations. that the solubility of the precipitate in the liquid present
cannot be very low. This solubility appeared to the authors
Table I-Effect of Varying Time of Standing on Canadian Lead greater than was t o be expected if the water present were the
Number
--SIRUP 2-- --SIRCP 11-- ---SIRUP 13-- sole solvent.
TIME
A B Mean A B Mean A B Mean
Table 111--Effect of Varying Amounts of Lead Subacetate on
Hours Canadian Lead Number
1/z 2 . 5 4 2.64 2.59 2 . 4 1 2.37 2.39 3.19 2.99 3.09 (Averages of duplicates)
1 2 . 7 9 2.78 2.79 2.35 2.41 2.38 3.20 3.34 3.27 VOLUME OB REAGENT SIRUP 3 SIRUP 1 SIRUP 2 SIRUP 4 SIRUP 7
ll/n 2.69 2 . 7 3 2 . 7 1 2.44 2.46 2.45 3.34 3.26 3.30
2 2.74 2.84 2.79 2.42 2.28 2.35 3.35 3.46 3.40 cc.
2'/1 2.78 2.90 2.84 2.38 2.38 3.31 3 . 1 0 3.20 0.5 .... 2.066 1 , 9 3 6 2 . 3 6 8 2 506
3 2 . 8 1 2 . 9 1 2 . 8 6 2 . 3 7 2 . 3 7 3 .03 3.14 3.08 1.0 2,508 2.633 2 . 7 1 0 3 074 4 136
2.24 2.28 2.26 3.16 3.29 3 . 2 2 1.5 2.470 2 , 5 9 4 2 , 7 2 7 3 . 1 5 2 4 658
2.0 2,408 2 , 5 7 7 2 , 7 4 4 2 . 9 6 2 5 042
2.5 2.250 2 . 4 5 3 2 , 5 9 4 2 . 8 5 2 4 916
Precipitation at Various Temperatures 3.0 2,132 2 . 1 5 0 2 . 3 4 3 2 , 7 2 8 4 928
4.0 1,593 1.674 1.978 2,442 4 722
5.0 .... 1.121 .... 1 . 9 1 8 4 294
The aliquot sample was heated to the required temperature 4 . 0 reduction from maximum 0.915 0.959 0.766 0.710 0.320
Per cent reduction of maximum 3 6 . 5 36.4 27.9 22.5 6.3
on a water bath, the subacetate added, and the temperature 4 . 0 reduction from 2 . 0 0.815 0 . 9 0 3 0 . 7 6 6 0 . 5 2 0 0 . 3 2 0
maintained within 2 degrees for 2 hours. Per cent reduction of 2 . 0 cc.
number 33.8 35.1 27.9 17.5 6.3
The results in Table I1 are the averages of duplicate de- 5 . 0 reduction from 2 , 0 .., . 1.456 . . . . 1,044 0.748
terminations. They are represented graphicaJly in Figure l. Per cent reduction of 2 . 0 cc.
At 100" C. the precipitates were light brown, coagulated im-
number . . . . 56.5 . . .. 35.2 14.8
Range of values in the five sirups:
mediately, and darkened to deep chocolate on standing. At Using 1 .O cc. reagent, 4 . 1 3 6 - 2 , 5 0 8 = 1 , 6 2 8 = 6 4 . 9 % of minimum
80" C. the precipitates curdled immediately; on standing, a
Using 1 . 5 cc. reagent, 4 . 6 5 8 - 2 , 4 7 0 = 2 . 1 8 8 = 8 8 . 6 Yoof minjmum
Using 2 . 0 cc. reagent, 5 . 0 4 2 - 2 , 4 0 8 = 2 . 6 3 4 = 1 0 9 . 5 %
' of minimum
clear sirupy layer formed a t the surface, in which a fine
secondary precipitate developed, similar to that formed a t Work on the Winton method had led RossI3 to the con-
room temperature. After 2 hours the color of the precipitate clusion that the precipitates produced by interaction of maple
resembled that of the room-temperature, rather than that of sirups and basic lead acetate solutions were soluble both in
the 100" C., precipitate. sucrose and in excess of the subacetate. As this would ac-
count both for the incomplete precipitation of lead by excess
Table 11-Effect of Temperature of Precipitation on Canadian Lead of sirup and for the falling off of lead values in presence of
Number
(Averages of duplicates)
excess of subacetate, it was decided t o seek confirmation of
TEMPERATURE
SIRUP 9 SIRUP 11 SIRUP 12 SIRUP 13 Ross's conclusion.
c.
100 2.826 1.590 2.174 2.386
80 3.284 1.512 2.240 2,500
60 3.620 1.539 2.139 2,800
40 4,484 2.230 2.696 3,600
20 4.630 2.640 2.840 3,676
in the usual way, averaged 2.672 for the sucrose duplicates lead numbers (1 and 3) give a greater precipitation with 1
and 3.786 for the water checks. than with 2 cc. It will be noted also that, in he five sirups
The results of both series of experiments evidence a con- involved, the range of variation of the lead number is much
siderably greater solubility of the precipitate in sucrose solu- narrower when 1 than when 2 cc. of reagent are used-vie.,
tion than in water. 65 per cent of the minimum of value as compared with 110
per cent. This point deserves study on a greater number of
Effect of Excess of Lead Subacetate samples. The fact that in sirups of hi,gh lead value the maxi-
Four aliquots from each of three sirups were precipitated in mum precipitation would not be reached with less than 2 cc.
the usual manner. After standing 1 hour-a time found of reagent is not important, as i t is those of low lead value
sufficient for complete precipitation (see Table 1)-an ad- that are likely t o be under suspicion.
~
ditional 3-cc. portion of the subacetate was added to each of Regulation of Washing
two of each set of four aliquots. All were then allowed to
stand 1 hour longer and the determinations completed as I n the current method filtration becomes slow as soon as
usual. The results, recorded in Table IV, show clearly that the precipitate is transferred to the Gooch, but as soon as
the precipitate (or some constituent thereof) is freely soluble all the filtrate has passed through, cracks form in the pre-
in excess of the reagent. Both Ross's conclusions are thus cipitate, and added wash water runs through very quickly.
confirmed. Hot water tends to cause the cracking, eveii when the pre-
The solvent effect of sucrose will probably not vary much cipitate is kept covered. As the surface of contact and the
from sirup to sirup. The same quantity of dry matter, as time of contact of the precipitate with the wash water vary
indicated by the refractometer, is used in all the determi- considerably from experiment to experiment, differences in
nations of Canadian lead value, and the proportions of duplicates are to be anticipated and the magnitude of these
invert sugar and other constituents in this dry matter are differences will depend upon the solubility of the lead-maple
small in comparison with the sucrose. The proportions of precipitate in hot water.
the various substances which react with the basic lead acetate
Table V-Solubility of Lead-Maple Precipitate in Hot Water
no doubt differ also to some extent in different sirups, and CANADIAN PRECIPITATBDISSOLVED
the solvent effect of the sucrose may vary to some extent with LEAD TOTALSOLIDS IN In 260 CC. In 500 CC.
SIRUP NUMBER 100 cc. ALIQUOTS water water
the composition of the precipitate. Inasmuch, however, as Gram Per cent Per cent
the range of variation of lead value in genuine sirups is de- 36 2.2 0.0173 38.05 58.08
termined by the same procedure as is used in testing for 41 3.1 0.0266 43.50 60.62
40 5.2 0.0373 34.12 60.49
adulteration, the solvent effect of the sucrose would not appear
to be a factor of much significance in reference to the detec- To get an idea of the magnitude of this solubility, the well-
tion of adulteration. Experimentation upon concentrations washed precipitates from the Canadian lead number deter-
other than 25 grams dry matter per 100 cc. might possibly be minations of three sirups together with the asbestos were
fruitful, but the prospect is not so good as for modifications of transferred to flasks, heated on a steam bath 5 hours with
the method in other directions. 250 cc. water under return condensers, and brought to boiling.
Table IV-Solvent Effect of Lead Subacetate on Canadian Lead
One hundred cubic centimeter aliquots of the filtered solution
Precipitates were evaporated a t 100" C. The results are shown in Table
(Averages of duplicates) V. Portions of the filtered solutions also were allowed to
CANADIANLEADNUMBERS
METHOD Sirup 36 Sirup 13 Sirup 40 cool. In these a fine white precipitate formed, showing that
Usual procedure with 2 cc. subacetate 2.216 3.324 5.280 the solubility is greater in hot than in cold water.
With addition of 3 cc. extra subacetate
after 1 hour 1.216 1.860 4.440 The residues from the first treatment were again treated
Reduction 1.000 1.464 0.840 with 250 cc. of water in the same manner, with the results
Per cent reduction of 2 cc. number 45.1 44.1 16.9 shown in the last column of Table V. It is clear, then, that
the solubility of the precipitate in hot water is great enough
As regards the basic lead acetate, it is very evident that the to demand careful regulation of the washing process.
solvent effect will not be identical for all sirups. The lower
Washing by Centrifugal Decantation
the content of precipitable matter, the greater will be the ex-
cess of basic acetate and, therefore, the greater its solvent It appeared that more uniform conditions of time and con-
effect on the precipitate. Consequently, low lead numbers tact ought to be obtainable by a decantation process. The
will be reduced much more than high ones. The range of slow settling of the precipitate, however, offered a difficulty,
the Canadian lead number in genuine maple. goods will there- and centrifugal settling was adopted. The precipitation was
fore be wider than the range of content of the substances upon performed in centrifuge tubes. After 2 hours' standing the
which the determination depends. (The data of Tables I11 tubes were whirled for a few seconds in a Babcock milk-test
and IV show a greater effect of excess of subacetate on low centrifuge. The supernatant liquid was decanted to the
than on high lead values.) In spite of this disadvantageous Gooch and 17.4 cc. of hot water were delivered into the tube
effect of the excess of reagent, however, the range of variation from an automatic pipet, connected by asbestos-insulated
of Canadian lead number appears to be narrower than that tubes with a flask of boiling water. After mixing and a few
of the majority of analytical values of maple products.14 I n seconds' standing, the tube was again whirled in the centri-
particular, it appears to be distinctly narrower than that of fuge and the supernatant liquid decanted. This was re-
the malic acid value-a value determined by a method (pre- peated to a total of four washings. After the second washing
cipitation by calcium and alcohol) designed to measure the the use of the centrifuge was usually superfluous as the pre-
constituent which is largely responsible for the precipitation cipitate had become granular and so settled well. A fifth
of the lead of the subacetate reagent, viz., the malate radical. and a sixth portion of wash water were used in transferring
The solvent effect of the lead subacetate might probably the precipitate to the Gooch, making 104.4 cc. of wash water
be diminished and the range of lead numbers in genuine goods in all.
somewhat narrowed by using a smaller quantity of the reagent The results of eight determinations by the established
than 2 cc. Table I11 shows that the two sirups with lowest method and six determinations by the centrifugal decantation
14 Trans. Roy. SOC.C a n , [3], 13,111, 221 (1919). method on each of three sirups are summarized in Table VI.
January 15, 1929 IilrDL-STRIAL AND ENGINEERING CHEMISTRY 11
It will be observed that washing by decantation reduces the tate in the hot water. Using the regular method in a similar
range of variation to about half that shown where ordinary series of experiments, Scotts failed to detect this difference
washing was practiced. The variations between duplicates and concluded that it was indifferent whether 100 or 150 cc.
in the writers determinations by the established Canadian of hot water were used in the washing.
lead method are of the same general magnitude as in those
reported by the Laboratory of the Inland Revenue Depart- Time Required for Complete Precipitation
ment.15 The latter were 0.10 in ten sirups of average lead Determinations by the decantation method were made
number 1.92 = 5.2 per cent; 0.13 in 10 sirups of average lead upon triplicate aliquots from each of three sirups with, re-
number 1.97 = 4.6 per cent; 0.16 (or, according to the figures spectively, l/6, I/*, 1, 2, and 5 hours standing. The results
of the table as printed, 0.19) in twenty-four sirups of average (Table IX) show a steady increase in lead number with time
lead number 3.11 = 5.1 (or 6.1) per cent. of standing. The original method failed to detect such a
Table VI-Comparison of Canadian Lead Numbers, Decantation time effect both in the writers hands (Table I) and in those of
Method with Ordinary Method Valin.10 This is to be attributed to the superior precision
SIRUP31 SIRUP44 SIRUP48
Variation Variation Variation
of the decantation method.
METHODO F ma:. to max: to ma:. to
Table IX-Effect of Time on Precipitation. Canadian Lead Num-
WASHINGPPT. Mean min. Mean min. Mean min.
bers-Decantation Method
Per cent Per cent Per cent (Means of triplicates)
of mean of mean oj mean
Decantation : SIRUP 10 MINUTES30 MINUTES 1 HOUR 2 HOURS 5 HOURS
6 detns. 2.353 2.61 3.682 3.04 4.721 1.44 35 2.581 2.608 2.685 2.752 2.778
In crucible: 36 1.960 2.024 2.076 2.080 2.147
8 detns. 2.107 5.79 3.472 5.88 4.325 2.68 25 3.324 3.467 3.496 3.496 3.577
Duplicate determinations made on nine sirups by the de- Washing with Cold Water without Centrifugal
cantation method gave the results reported in Table VII. Decantation
It will be observed that the average variation of duplicates It was found that cracking of the precipitate in the Gooch
amounts to only 0.027 or 0.8 per cent of the mean, and the crucible could be prevented by washing with cold water,
maximum differences to 0.056, equal to 1.9 per cent of the
mean. taking care to keep the precipitate continually covered. (It
has been observed that when the rate of filtration exceeds
Table VII-Canadian Lead Numbers-Decantation Method 40 drops per minute, cracking $may occur even under cold
DIFFERENCEDIFFEBENCE
SIRUP A B MEAN OF DUPLICATESIN MEAN water.-J. F. S.)
Per cent Eight aliquots from each of three sirups were treated by
31 2.336 2.360 2,348 0.024 1.0 this method and eight by the regular method. In the cold
36 2.296 2.280 2.288 0.016 0.8
38 4.132 4.100 4.116 0.032 0.7 washing the crucibles were filled four times with cold water,
40 5.680 5.716 5.698 0.036 0.6 each portion being added just before the last of the preceding
44 3.632 3.682 3.657 0.050 1.4
47
48
3.544
4.608
3.528
4.600
3.536
4.604
0.016
0.008
0.45
0.2
liquid passed through. The results are summarized in Table
49 2.936 2,992 2.964 0.056 1.9 X. They show ,marked superiority for the cold water method
50 2.688 2.692 2.690 0.004 0.15
Average 3.545 0.027 0.8
as regards precision. It may be pointed out that greater
variations among multiplicate determinations by the regular
In all the nine sirups of Table VI1 higher results were method are shown with sirup 42 than with those included in
obtained by the decantation method than by the ordinary Tables VI and VII, and that in this series of experiments on
Canadian method. This was unexpected, but may be due sirup 48 a considerably wider range (though the same mean)
to the facts (1) that the wash water after passing through was found than in the series reported in Table VI.
the automatic pipet has a lower temperature than when Table X-Comparison of Canadian Lead Numbers Cold Water
poured directly from a wash bottle into the crucible, and Method of Washing with Ordinary Methbd
(2) that it cools still further during the centrifugation. I n SIRUP42 SIRUP43 SIRUP48
Variation Variation Variation
other words, though more thorough contact of precipitate METHODO F max. t o max. to max: to
WASHINQ Mean min. Mean min. Mean min.
with wash water is secured than in the regular method, that Per cent Per cent Per cent
contact is with water of lower solvent power than in the of mean of mean of mean
regular method. It would seem probable that centrifugal Cold water:
8 detns. 3.310 4.11 2.933 3.55 4.495 2.49
washing with water of a temperature equal t o that of the H o t water:
water used in the regular method would give lower, rather 8detns. 3.180 11.07 2.710 4.30 4.329 8.59
than higher, lead values; but such washing is obviously From the eight determinations on sirup 48 by the ordinary
difficult to secure in practice. method, reported in Table X, the lowest value obtained was
Volume of Wash Water 4.152 and the next lowest 4.252 (which coincides with the
minimum reported for the same sirup by the same method
Sets of three aliquots of solutions of two sirups were treated in Table VI). If we reject the lowest value, the mean be-
by the decantation method with, respectively, 3, 6, 9, and 12 comes 4.354 and the range (4.252-4.524 = 0.272) 6.25 per
portions of hot water from the 17.4-cc. pipet, making the cent of the mean, which is still considerably more than twice
approximate total volumes of wash water 50, 100, 150, and that of the values by the cold washing method.
200 cc. The results are recorded in Table VIII. , Compared with washing with hot water by centrifugal de-
Table VIII-Effect of Volume of Wash Water on Precipitate. Cana-
cantation, washing with cold water appears to be somewhat
dian Lead Numbers-Decantation Method less precise but more convenient. I n the sirups to which it
[Means of tridicates) has been applied it has, like the decantation method, yielded
Volume, cc. 50 100 150 200 lead values somewhat greater than those by the regular
Sirup 38 4.320 4,026 3.874 3.769
Sirup 54 2.496 2.279 2.133 1.843 method.
It should be pointed out that the ranges of variation dis-
The steady decrease of lead number with increasing amount cussed here are those between determinations made upon
of wash water is attributable to the dissolving of the precipi- aliquots of the same dilution of a sirup by the same observer
16 Laboratory Inland Revenue Dept. (Ottawa), Bull. 228, 11 (1911). on the same day. While, therefore, well adapted to their
12 ANALYTICAL EDI TIOW Vol. 1, y o . 1
purpose, they no doubt represent a degree of precision much covered with water and filtering a t a rate not exceeding 40
beyond that to be expected of results by different observers drops per minute. Dry a t 100" C., weigh, and multiply the
or even by the same observer a t different times. In Table weight by 20.
VII, for example, we have two determinations on sirup 48 by
the decantation method which both give values far below the Summary
minimum found by the same observer on another day in the 1-In the current Canadian lead method, large experi-
series reported in Table VI. mental errors occur due to the formation of fissures in the
Proposed Modification of Method precipitate and consequent irregularities in the solvent action
of the hot wash water.
Further study of the advisability of using 1.0 or 1.5 cc. of 2-Washing by decantation, using centrifugal settling,
subacetate reagent is being made under direction of the senior eliminates the error due to cracking and greatly reduces the
author (J. F. S.). Retaining for the present the use of 2 cc., difference between duplicates. With the greater precision
the directions for the cold washing method (Canadian lead attained by this procedure, it becomes evident that increasing
method, Fowler modification) are as follows: the volume of wash water decreases the lead number.
PREPARATION OF LEADSUBACETATE SoLuTIon--Boil 280 3-The lead-maple precipitate is soluble in sucrose solution
grams of Horne's dry basic lead acetate with 500 cc. of water. and in excess of the basic lead acetate solution. In different
Pour off from slight undissolved residue, allow to cool, and sirups maximum precipitation is realized with different quan-
dilute with recently boiled water to a density of 1.25 a t 20" C. tities of the reagent.
DETER?rIIxATIos-Weigh the quantity of prepared sirup 4-Much lower results are obtained by precipitating a t
containing 25 grams of dry matter, transfer to a 100-cc. 40" C. than a t room temperature. Results a t BO", 80, and
volumetric flask, and make up to mark a t 20" C. Pipet 20 100" C. are lower than at 40" C.
cc. into a large test tube, add 2 cc. basic lead acetate solution, 5-The amount of precipitate increases slightly with the
cork, and allow to stand 2 hours. interval between addition of reagent and filtration.
Filter with suction on a 25-cc. tared Gooch having an as- 6-Washing with cold water, keeping the precipitate cov-
bestos mat a t least 3 mm. thick. Before quite all the liquid ered therewith, and avoiding too rapid filtration, prevents
has passed through, fill the crucible with cold water. Repeat cracking and gives results much more precise and somewhat
the washing to a total of four times, taking care t o prevent higher than those obtained in the current procedure. The
formation of fissures by keeping the precipitate completely adoption of cold water washing is proposed.
18 Assocn. OfficialAgr. Chem , iMethods, 1925, p 204. were made on precipitates from each sirup. The precipitate
17 Snell, J. IND. ENG.CHEM.,8, 144 (1916). and asbestos were transferred to a combustion boat and
January 15, 1929 I N D LTSTRIAL A,VD EhTGIA7EERIhTGCHEMIXTRY 13
burned in oxygen in an electrically heated combustion fur- The wide variations in the quantities of total lead pre-
nace in the usual manner, a copper spiral being used. By cipitated by a given quantity of dry matter of sirup are at-
use of blank determinations the hydrogen results were cor- tributable, a t least in part, t o tha,t tendency of the precipi-
rected for water given off by the asbestos. The residue in tate to dissolve in excess of reagent discussed in Part X.
the boat contained metallic lead and litharge. What chemical action is involved in such dissolving is not
M A i x ACID-TO determine generally what proportion of known.
the tolal organic acids in the precipitate is malic, about fifty The two experiments on the combined precipitates from
precipitates, obtained from a large (unrecorded) number of miscellaneous sirups using diff ereiit decolorizing carbons give
sirups, mere ground together and transformed with hydrogen results indicating, respectively,that the malic acid constituted
sulfide as described above. The solution of the acids (1500 95.93 and 98.08 per cent of the total acids liberated by hydro-
cc.) was concentrated to about 200 cc., boiled with decoloriz- gen sulfide.19 The tabulated figures for normal lead malate
ing carbon, and filtered. Different decolorizing carbons were are therefore the equivalents of 97 per cent of the acids so
used in duplicate experiments. liberated. Added to the lead oxide as base, these account for
Total acids were determined in a measured volume of the only 83 to 84 per cent of the weight of the precipitate. The
filtrate by titration with 0.05 11- sodium hydroxide. Malic carbon of these quantities of lead malate is little more than
acid was determined by polarization of the neutralized solu- half that actually found in t'he combustion analysis. Even
tion, with and without saturation with uranyl acetate.18 were we t o assume the neutral lead as determined by differ-
ence bet'weeii total and basic lead to be all present as lead
Discussion of Results malate, the carbon of this salt would be considerably short of
that found by combustion. Similarly, the hydrogen of the
The results are summarized in the accompanying table. malate, calculated either way, falls far short of that found by
The lclad content of precipitates is higher, the lower the combustion. It is to be inferred that, in addition to basic
Canadian lead number. For two of the sirups the results on lead malate, the precipitate contains a considerable quantity
total lead are higher than any of Scott's; for the third, lower of highly carbonaceous organic matter.
than a n y but one of Scott's. It should be borne in mind, I n the transformation of the lead precipitate with hydrogen
however, that the method of washing and also the method of sulfide it was observed t'ha,t the acid filtrate was colorless.
determining the lead were different in the two investigations. On neutralization a brown color developed. When the
The basic lead is higher, the lower the Canadian lead number. neutralized, but not when the acid, solution was concentrated,
maple odor could be detected. The flavoring and coloring
C o m p o s i t i o n of C a n a d i a n L e a d P r e c i p i t a t e s of T h r e e M a p l e S i r u p s matters of the sirup appear, therefore, to be contained in the
SIRUP36 SIRUP41 SIRLP40
Run of sap Early Middle Middle
lead precipit'ate.
Evaporator ,Modern Iron Modern Sitrogen and sulfur were overlooked in the ult'imate analy-
Color 5 8 19
Canadian lead value 2 19 3 09 5 15 sis, Previous investigations20 suggest that these may possibly
Total carbon, per cent 11 13 11 7 1 12 23 constitute 1 or 2 per cent of the precipitate, provided the
Total hydrogen, per cent 1 46 1 22 1 15
Oxygen, nitrogen, sulfur, e t c , by dif- lead subacetate precipitates all or nearly all the compounds
ierence
Total lead, per cent
14 77
72 64
16 54
70 53
19 04
67 58
containing them. I n further work on this line the writers
100 00 100 00 100 00 would suggest t'he determination of these elements and also
Basic lead, per cent 38.62 38.53 35.44 that of reducing sugars in the filtrate from the hydrogen
Neutral lead, by difference 34.02 32.00 32.14 sulfide treatment.
Seutral lead equivalent of organic acids 25.95 26.70 2S,93
Neutral lead in excess of organic acids 8.07 5.30 3.21
Ratio neutral lead (by acids) t o basic lead 0.77 0,69 0.82 Summary
Ratio of neutral lead (by difference) t o
basic lead 0.88 0.83 0.91
Total lead precipitated by 100 grams dry 1-The Canadian lead precipitates of three maple sirup
matter of sirup, grams 1.65 2.24 3.68 samples have been analyzed for t,otal and basic lead, carbon
Normal lead malate equivalent of 97 per
cent oi the organic acids 41.23 42.40 45.99
and hydrogen, and total (organic) acids, liberated by hydro-
Lead oxide as base 41.61 41.51 38.18 gen sulfide.
Sum 82.84 83.91 84.17 2-The total lead content is greater, the lower the Canadian
Carbon of normal lead malate as esti- lead value, varying from 72.64 t o 67.58 per cent.
mated above 5.84 5.99 6,5l
Carbon of lead malate, per cent of total 3-The basic lead content appears t o be slightly higher,
carbon 52 5 51 1 53 2 the lower the Canadian lead value.
4-The difference between the tot.al and basic lead con-
Calculated as malic acid the percentages of organic acids siderably exceeds the lead equivalent of the acids liberated
are: ;,irup 36, 16.56; 41, 17.02; 40, 18.46. In the table the by hydrogen sulfide.
equivalent amounts of neutral lead are given. As thus cal- &-The inalic acid, as determined polarimetrically in a
culated, the neutral lead is lower, the lower the Canadian lead composite precipitate from more than twenty sirups, con-
d u e . There is, however, a considerable discrepancy be- stitutes 97 per cent of the acids liberated by hydrogen sulfide.
tween this estimate of neutral lead and that obtained by 6-The precipitates were found t o contain 11.1t o 12.2 per
taking the difference between total and basic lead, the latter cent of carbon and 1.46 to 1.15 per cent of hydrogen. This
being always the greater. This suggests that some of the indicates that substances much more carbonaceous than lead
neutral lead of the precipitate may be associated with sub- malate are present.
stances which are not determinable as acids by titration. 7-The flavoring and coloring matters appear to be pre-
The percentage of neutral lead by difference is nearly equal cipitated by the basic lead acetate.
in the precipitates from the two sirups of higher lead values,
18 Nelson, J . Am. Chem. Soc., 60, 2006 (1928), finds material quantities
but distinctly greater in the other sirup (36). The content of of formic, acetic, and citric acid in maple sirup. The last mentioned was
basic lead is decidedly higher than that of neutral lead, indi- discovered by him in a lead subacetate precipitate. Hls estimate makes i t
cating the presence of salts more basic than the lead malates constitute S t o 12 per cent of the sum of the malic and citric in the sirup.
mentioned in the introduction. Oui analyses would allow of only 2 t o 4 per cent of the acids of the precipi-
tate being other than malic.
13 Assocn. Official Agr. Chem., Methods, 1925, p. 213, Dunbar and 20 Hortvet, J . Am. Chem. Soc.. 26, 1523 (1904); Snell, Trans. Roy.
Bacon, U. S. Dept. Agr., Bur. Chem. Czrc 76 (1911). SOC.Can., [3] 13, 111, 221 (1919).
14 ANALYTICAL EDITION Vol. 1, No. 1
RECENT paper from this laboratory2 described electrodes and a continuous record of the conductivity can
A a continuous method for determining small concentra- be obtained by means of a Leeds and Northrup recording
tions of sulfur dioxide in air, which is automatic to the Wheatstone bridge. The bridge has a range from 125 t o
point a t which the solutions are discharged from the machine 40,000 ohms. The cell constant is about 0.06 reciprocal
ready for titration. In order to make the method completely centimeter. Temperature compensation has been accom-
automatic, it is necessary to measure the change in some plished during the past season by a rheostat and scale on the
property of the absorbing solutiqn by a recording device. recorder, because a compensating cell sufficiently constant
The absorbent employed, containing 0.2 per cent potassium for use with the dilute solutions employed was not available.
iodide and 0.1 per cent starch together with the necessary The writers propose to construct such a cell. T o obtain a
iodine, offers serious difficulties to the accomplishment of complete record of each absorption a recorder for each ab-
this purpose. The intensity of starch-iodine color might sorber is necessary, but if a part of the record can be omitted,
be measured, but the range of iodine concentration which is or if the ga,s sampling is interrupted for 20 to 30 seconds after
accompanied by an appreci- the completion of an absorp-
able color change is too small tion, to allow the recorder
to be useful. Since the ab- The use of slightly acidulated distilled water contain- time' to return to its initial
sorption of sulfur dioxide ing 0.003 to 0.006 per cent hydrogen peroxide is proposed p o s i t i o n , a s w i t c h can
produces one e q u i v a l e n t as an absorbent for traces of sulfur dioxide in air to re- readily be installed so that
each of sulfuric acid and hy- place the iodine solution previously used for this pur- one recorder will serve both
driodic acid, it was thought pose. The absorption of the gas in the peroxide solu- absorbers.
that if the potassium iodide tion can be followed either by titration with alkali using
Preparation of Reagent
i n t h e s o l u t i o n could be methyl red as an indicator or by a conductivity method.
e l i m i n a t e d a conductivity A study of the method shows that it is fully as reliable The reagent consists of
method or an iodine elec- as the iodine method in the absence of sulfur trioxide, distilled water to which is
trode could be employed. and permits the construction of apparatus which is added 0.1 to 0.2 cc. 30 per
It was impossible, however, completely automatic. cent hydrogen peroxide and
to omit the salt from the ab- 0.3 cc. 0.1 N sulfuric acid
s o r b e n t without rendering per liter. The solution is
the iodine quite volatile. iimilarly, it was not practicable to aspirated for a few minutes with pure air to adjust the carbon
measure the hydrogen-ion concentration because of the reac- dioxide equilibrium. One hundred cubic centimeters of this
tion between iodine and both the hydrogen electrode and the solution will take care of 2 gram-molecular volumes of air
quinhydrone electrode. containing 50 to 100 p. p. m. sulfur dioxide. The recorder
is calibrated by adding known amounts of sulfuric acid to the
Hydrogen Peroxide as Absorbent
solutions in the absorbers and aspirating with pure air, as in
Because of these difficulties another absorbent was sought. the regular operation of the machine.
I was observed that distilled water is an excellent absorbent The addition of the acid in the preparation of the reagent
for appreciable amounts of the sulfur dioxide. For example, serves two purposes. Carbonates are decomposed,,permitting
a t 5 p. p. in. practically all of the sulfur dioxide can be removed a rapid adjustment of the carbon dioxide equilibrium. This
from 30 liters of air aspirated a t 15 liters per minute. At is essential because the recorder may show an increased
50 p. p. m. only4 per cent of the gas reaches a second absorber resistance when the absorption commences, corresponding
under these conditions, but about 25 per cent of the absorbed to an error as great as 0.5 p. p. m. sulfur dioxide if distilled
material is oxidized. For analytical purposes it is, of course, water which has been standing in glass bottles for a week or
necessary either to prevent this oxidation or carry it to com- two is used without this preliminary treatment. If the
pletion. The second course is easily accomplished by the concentration of the sulfur dioxide in the sample is constant,
use of hydrogen peroxide. This is an ideal oxidant, because it is possible to correct the recorder reading for this effect
it is a non-electrolyte which reacts rapidly with sulfur dioxide. by noting the slope of the absorption curve and the time of
Thus the liquid remains an efficient absorbent for larger sampling. The acid also reduces the resistance of the soh-
amounts of the gas, and the conductivity of the solution can tion to less than 8000 ohms. At higher resistances the
be employed for quantitative measurements. recorder is somewhat insensitive to added acid. I n the
range between 8000 and 1500 ohms, representing 0 to 5 p. p. m.
Modifications in Apparatus sulfur dioxide under the conditions of operation, the instru-
I n applying the results of this preliminary work only slight ment responds linearly a t the rate of 0.2 p. p. m. sulfur dioxide
modifications of the apparatus previously described had to per scale division. At lower resistances the response of the
be made. The discharge valve of each absorber is placed on recorder gradually diminishes, until a t 150 ohms, representing
the outside of the vessel and the platinum electrodes, mounted about 50 p. p. m. sulfur dioxide, the rate is only 2 p. p. m. per
on the end of a glass tube, are inserted near the bottom of the scale division.
abfiorber about 1 cm. below the circular bubbler. In this Titration
position the gas bubbles do not come in contact with the
1 Received November 19, 1928.
As a check on the recorder readings the aspirated solutions
2 Thomas and Cross, IND. ENG.CHEM., 20, 645 (1928). may be titrated with sodium hydroxide using methyl red a s
January 15, 1929 IhTDUSTRIAL A N D ENGINEERING CHEMISTRY 15
an indicator. A definite amount of the indicator-for ex- second absorber placed in series with the first having indicated
ample, 0.5 ec. of 0.02 per cent solution, the reaction of which that this amount of gas escaped absorption. At 15 p. p. m.
has been adjusted to approximate neutrality-is added to less than 1 per cent of the gas reached a second absorber.
the solution before titration and the end point determined
by matching with the faint pink color of a standard solution. Advantages of New Method
This titration is very readily carried out and the end point The new method has several advantages over the iodine
is extremely permanent and sensitive, 0.1 cc. of 0.002 N alkali method. The solutions are easier to prepare and the ab-
being sufficient to cause a definite color change. A blank sorption is as rapid and complete as with the iodine method.
titration must also be made. It is not necessary to adjust the strength of the oxidant to
Table I-Sulfur Dioxide in Mixtures of Sulfur Dioxide a n d Air a s In-
correspond to the particular concentrations of sulfur dioxide
dicated by Hydrogen Peroxide a n d Iodine Methods with which it is desired to work. The data give the amount
of gas absorbed directly and not indirectly through the de-
sox CONCENTRATION DIFFERENCE BETWEEN CON-
DUCTIVITY METHOD AND: crease in concentration of another constituent of the system.
I
IODINE METHOD
Conduc- Acid TION METHOD
- Measurement of Both SOz and SOs
tivity titration Iodine Aver- Maxi- Aver- Maxi-
~ age mum age mum If it is desired to measure both sulfur dioxide and sulfur
trioxide in a gas mixture, the two methods could be con-
veniently applied to this end-the iodine method to measure
1 0.37 0.35 0.38
1 0.39 0.37 0.38 the sulfur dioxide and the hydrogen peroxide to measure the
1 0.40 0.41 0.39 total acidity. As a general rule, however, the amount of
1 0.69 0.71 0.69
1 0.93 0.90 0.88 sulfur trioxide is inappreciable compared with the sulfur
6 0.35 0.30 0.35 0.05 0.10 0.00 0.03
15 1.36 ... 1.44 0.07 0.10 dioxide, and can be neglected.
20 1.71 1.61 1.75 0: i7 0: 40 0.04 0.09
20 3.19 2.90 3.24 0.30 0.50 0.06 0.10 Application in the Field
6 6.41 6.40 6.43 0.07 0.10 0.14 0.25
15 13.58 13.55 13.40 0.20 0.50 0.27 0.5
8 24.1 24.3 24.4 0.2 0.3 0.3 0.7 The method can doubtless be applied to the problem of
12 36.4 37.2 35.0 0.8 1.7 1.4 2.2
4 58.3 59.1 56.4 0.8 1.2 1.9 2.7 the continuous measurement of the concentration of sulfur
-- dioxide in the field. Only slight modifications would be
necessary to permit the apparatus to operate for 1 or 2 days
Comparison of Methods
without any attention. The period of sampling could con-
The new method has been checked against the iodine veniently be increased to about 10 to 20 minutes since con-
method, previously described, by analyzing uniform gas centrations greater than 5 p. p. m. would seldom be encoun-
mixtures simultaneously with both procedures. The results tered. The final adjustment of the carbon dioxide equilibrium,
are given in Table I and ,show a satisfactory concordance over which is always accomplished in less than half a minute,
a range from less than 0.4 p. p. m. to 60 p. p. m. The five single would then have a negligible effect on the results. It would
det,erminations represent large samples of gas, 120 t o 200 also be desirable to have a record of the gas volumes and
liters; the other values are the mean of a number of con- liquid volumes, even though these values are constant. A
secutive analyses of a uniform gas mixture carried out accord- thermograph and barograph would complete the necessary
ing to the regular procedure. At the highest concentrations equipment to permit the calculation of the sulfur dioxide
the results are probably low to the extent of 2.5 per cent, a concentrations.
ONTINUOUS extractors for liquids with volatile im- sample in the bulb, up in
C miscible solvents have been described in the literature,2
but most of them'require either elaborate glass blowing,
or else, since fairly large volumes of solution are used, the extrac-
( A ) or down in ( B ) ,and
overflows back into the
receiver.
tion is very slow. If the extract can be
A simple apparatus was devised for the analytical extraction titrated directly, as in
of liquids. Two forms are illustrated ( A ) for solvents lighter the case of acetic acid,
than the extracted solution, and ( B )for heavy solvents. Each is it is convenient to pro-
made from a pipet of any desired size with the stems bent and vide a buret and follow
connections sealed, as shown. In ( A ) the bulb of the pipet is the extraction as it pro-
filled four-fifths full with glass beads before bending the upper ceeds. This permits ex-
stem. The bulb in ( B ) can be filled completely afterwards. trapolation to infinite
The sample, about half the capacity of the bulb, is pipetted time with great ac-
into the bulb, before the solvent in ( A ) and after it in ( B ) . curacy. An extraction
Enough solvent is used to fill the rest of the bulb and tubes and of acetic acid from 50 cc.
have 20 to 30 cc. in the receiving flask. The solvent is distilled of salt solution with
by means of a hot plate through the upper tube which is bent ether removed half the
downward to avoid excessive r d u x condensation. After drop- total in 8 minutes, half
ping back from the condenser, the solvent bubbles through the the remainder in the
1 Received November 9, 1928. next 8 minutes, etc., so
* Fiske, A m . Chem. J., 41, 510 (1909): Berlin, German Patent 251,459 that in an hour the ex-
(June 24, 1911), see C. A., 7 , 7 (1913);Watkins, IND. ENG. CHEM.,il, traction was practically b-----d
612 (1925). complete. A
January 15, 1929 IhTDUSTRIAL A N D ENGINEERING CHEMISTRY 15
an indicator. A definite amount of the indicator-for ex- second absorber placed in series with the first having indicated
ample, 0.5 ec. of 0.02 per cent solution, the reaction of which that this amount of gas escaped absorption. At 15 p. p. m.
has been adjusted to approximate neutrality-is added to less than 1 per cent of the gas reached a second absorber.
the solution before titration and the end point determined
by matching with the faint pink color of a standard solution. Advantages of New Method
This titration is very readily carried out and the end point The new method has several advantages over the iodine
is extremely permanent and sensitive, 0.1 cc. of 0.002 N alkali method. The solutions are easier to prepare and the ab-
being sufficient to cause a definite color change. A blank sorption is as rapid and complete as with the iodine method.
titration must also be made. It is not necessary to adjust the strength of the oxidant to
Table I-Sulfur Dioxide in Mixtures of Sulfur Dioxide a n d Air a s In-
correspond to the particular concentrations of sulfur dioxide
dicated by Hydrogen Peroxide a n d Iodine Methods with which it is desired to work. The data give the amount
of gas absorbed directly and not indirectly through the de-
sox CONCENTRATION DIFFERENCE BETWEEN CON-
DUCTIVITY METHOD AND: crease in concentration of another constituent of the system.
I
IODINE METHOD
Conduc- Acid TION METHOD
- Measurement of Both SOz a n d SOs
tivity titration Iodine Aver- Maxi- Aver- Maxi-
~ age mum age mum If it is desired to measure both sulfur dioxide and sulfur
trioxide in a gas mixture, the two methods could be con-
veniently applied to this end-the iodine method to measure
1 0.37 0.35 0.38
1 0.39 0.37 0.38 the sulfur dioxide and the hydrogen peroxide to measure the
1 0.40 0.41 0.39 total acidity. As a general rule, however, the amount of
1 0.69 0.71 0.69
1 0.93 0.90 0.88 sulfur trioxide is inappreciable compared with the sulfur
6 0.35 0.30 0.35 0.05 0.10 0.00 0.03
15 1.36 ... 1.44 0.07 0.10 dioxide, and can be neglected.
20 1.71 1.61 1.75 0: i7 0: 40 0.04 0.09
20 3.19 2.90 3.24 0.30 0.50 0.06 0.10 Application in the Field
6 6.41 6.40 6.43 0.07 0.10 0.14 0.25
15 13.58 13.55 13.40 0.20 0.50 0.27 0.5
8 24.1 24.3 24.4 0.2 0.3 0.3 0.7 The method can doubtless be applied to the problem of
12 36.4 37.2 35.0 0.8 1.7 1.4 2.2
4 58.3 59.1 56.4 0.8 1.2 1.9 2.7 the continuous measurement of the concentration of sulfur
-- dioxide in the field. Only slight modifications would be
necessary to permit the apparatus to operate for 1 or 2 days
Comparison of Methods
without any attention. The period of sampling could con-
The new method has been checked against the iodine veniently be increased to about 10 to 20 minutes since con-
method, previously described, by analyzing uniform gas centrations greater than 5 p. p. m. would seldom be encoun-
mixtures simultaneously with both procedures. The results tered. The final adjustment of the carbon dioxide equilibrium,
are given in Table I and ,show a satisfactory concordance over which is always accomplished in less than half a minute,
a range from less than 0.4 p. p. m. to 60 p. p. m. The five single would then have a negligible effect on the results. It would
det,erminations represent large samples of gas, 120 t o 200 also be desirable to have a record of the gas volumes and
liters; the other values are the mean of a number of con- liquid volumes, even though these values are constant. A
secutive analyses of a uniform gas mixture carried out accord- thermograph and barograph would complete the necessary
ing to the regular procedure. At the highest concentrations equipment to permit the calculation of the sulfur dioxide
the results are probably low to the extent of 2.5 per cent, a concentrations.
ONTINUOUS extractors for liquids with volatile im- sample in the bulb, up in
C miscible solvents have been described in the literature,2
but most of them'require either elaborate glass blowing,
or else, since fairly large volumes of solution are used, the extrac-
( A ) or down in ( B ) ,and
overflows back into the
receiver.
tion is very slow. If the extract can be
A simple apparatus was devised for the analytical extraction titrated directly, as in
of liquids. Two forms are illustrated ( A ) for solvents lighter the case of acetic acid,
than the extracted solution, and ( B )for heavy solvents. Each is it is convenient to pro-
made from a pipet of any desired size with the stems bent and vide a buret and follow
connections sealed, as shown. In ( A ) the bulb of the pipet is the extraction as it pro-
filled four-fifths full with glass beads before bending the upper ceeds. This permits ex-
stem. The bulb in ( B ) can be filled completely afterwards. trapolation to infinite
The sample, about half the capacity of the bulb, is pipetted time with great ac-
into the bulb, before the solvent in ( A ) and after it in ( B ) . curacy. An extraction
Enough solvent is used to fill the rest of the bulb and tubes and of acetic acid from 50 cc.
have 20 to 30 cc. in the receiving flask. The solvent is distilled of salt solution with
by means of a hot plate through the upper tube which is bent ether removed half the
downward to avoid excessive r d u x condensation. After drop- total in 8 minutes, half
ping back from the condenser, the solvent bubbles through the the remainder in the
1 Received November 9, 1928. next 8 minutes, etc., so
* Fiske, A m . Chem. J., 41, 510 (1909): Berlin, German Patent 251,459 that in an hour the ex-
(June 24, 1911), see C. A., 7 , 7 (1913);Watkins, IND. ENG. CHEM.,il, traction was practically b-----d
612 (1925). complete. A
16 ANALYTICAL EDITIOhT Vol. 1. So. 1
HE writer has had occasion t o analyze organic liquids of nature, has great absorbent capacity for organic liquids and
TLCANIZED r u b b e r This test, by use of a sample of new design, subjects the jaws of the testing ma-
V is frequently required
to withstand, not only
the simple stresses such as
rubber to a combination of tensile and shearing
stresses. Shear, however, is the predominating stress.
Certain aged inner tubes have been found which
chine (Figure l). The lower
jaw moves a t the rate of 10
inches (25.4 cm.) per minute.
those of compression, tensile, deteriorated more when examined by this test than a I n the usual manner read-
and shear, but also the com- comparison of their tensile-stress-strain curves with ings of load and elongation
bined effects such as torsion, those of fresh tubes would indicate. Tearing action are taken up t o and includ-
tearing, bending, etc. It has seems to be approximated by this test. The effect of ing those at rupture. From
been observed that, as some overcure in some cases has been recorded at earlier these measurements and the
vulcanized rubber s a m p l e s stages by this test than by the tensile criterion. The cross section of the tongue the
age, their resistance to shear- test is easy to perform and with usual precautions shear-stress-strain curve may
ing and tearing stresses de- should have an accuracy of approximately 10 per cent. be plotted and the relative
creases inuch faster than their energy to start rupture esti-
resistance to tensile stresses, as determined under the stand- mated by determining the area under the curve. Figure 3
ard procedure of rubber-testing. shows the comparison of the shear-stress-strain curves with
During a study of natural and artificial aging some inner- the usual tensile-stress-strain curves. The reasons that the
tube samples, which tested very poorly after the oxygen bomb shear-stress-strain curves do not exactly coincide with the
(50 hours a t 60 C. and 20.4 atm.), were filed for further tensile-stress-strain curves seem to be: (1) Different widths
observation. As these tubes aged the tensile tests showed of samples were stretched; (2) different rates of stretching
less deterioration than was expected. However, a close were employed; (3) near break certain stocks seem to yield
examination of these tubes showed that they had developed some on shearing. In other words, the rupture is very slow
a very poor resistance to tear and were weak when sudden while in other cases it is instantaneous. These curves are
tensile stress was applied. For sake of brevity this lack of compared more fully in Table 111.
resistance t o sudden stress will be called shortness. A shorter and slightly less accurate method of getting the
After several attempts to measure this shortness property relative energy t o rupture the tongue is to take one-half the
without resorting to some new testing machine, the tongue product, S X E/100, where S is the stress a t rupture (kg. per
shear test was developed. It is so named because of the sq. cm.) and E is the per cent ultimate elongation. The
shape of the test specimen and the effect that is produced. product, the shear product, is close enough for ordinary
It seems to give about the proper rating to these short comparisons and closer to the real energy values than the
tubes. tensile product is for the relation that it expresses, be-
cause lower concavity factors exist in the shear-stress-strain
The Test curves.
A specimen like that in Figure 1 is prepared by means of Mechanism of Test
a cutting die, as shown by the pattern in Figure 2. Two
parallel marks exactly 1 inch (2.54 cm.) apart are placed on The manner of rupture and factors influencing the values
the test piece so that they will be near the middle of the obtained were studied to learn, first, what was taking place
parallel section of its tongue. The sample is then placed in as the rubber was strained in this sort of specimen, and second,
1 Presented before the Division of Rubber Chemistry at the 76th
if the test piece was properly proportioned. The rupture
Meeting of the American Chemical Society, Swampscott, Mass , September always takes place a t the end of the tongue (C-D, Figure 1).
10 to 14, 1928. A much smaller expenditure of energy is required to start
January 15, 1929 INDUSTRIAL AhrD ENGINEERIiVG CHEMISTRY 17
Beckmann* suggested that the low results in this particular KEW DEVICE CAPILLARYMETHOD
Carbon Hydrogen Carbon Hydrogen
case were due not only to incomplete comkiustion but to the Found Calcd. Found Calcd. Found Calcd. Found Calcd.
formation of pyromucates. which they found difficult tp burn 70 70 % % % 70 % %
completely. They found that, in spite of all precautions,
the analyses for carbon, in the case of the esters of n-amyl.
n-hexyl, n-heptyl, n-octyl of furoic acid, came out uniformly
low, especially in the case of the higher esters, and also in Analysis of a Known Highly Volatile Liquid
the case of the esters of furoylacetic acid.Q With the use of
this new device, however, the analysis of esters of similar As a final test two analyses were made of a highly volatile
nature, such as of a-tetrahydrofurfuryl alcohol synthesized by organic liquid compound, benzene, with the following results:
Zanetti,lo the author obtained very satisfactory results. The CARBON HYDROGEN
most difficult to burn in this series was found to be the Found Calcd. Found Calcd.
isovalerianate of a-tetrahydrofurfuryl alcohol, for which the 70 70 70 70
following analyses (not yet published by authors) were ob- 91.97 92,30 7.73 7.69
92.06 7.88
tained :
~-
* Zanetti and Beckmann, J . A m Chem. S O ~ .48,
, 1067 (1926).
These results indicate that volatile liquids like benzene can
9 Ibid., SO, 1438 (1928). be weighed in this new device and introduced into the com-
1 OZanetti, Ibid., 60, 1821 (1928). bustion tube without any material loss.
TLCANIZED r u b b e r This test, by use of a sample of new design, subjects the jaws of the testing ma-
V is frequently required
to withstand, not only
the simple stresses such as
rubber to a combination of tensile and shearing
stresses. Shear, however, is the predominating stress.
Certain aged inner tubes have been found which
chine (Figure l). The lower
jaw moves a t the rate of 10
inches (25.4 cm.) per minute.
those of compression, tensile, deteriorated more when examined by this test than a I n the usual manner read-
and shear, but also the com- comparison of their tensile-stress-strain curves with ings of load and elongation
bined effects such as torsion, those of fresh tubes would indicate. Tearing action are taken up t o and includ-
tearing, bending, etc. It has seems to be approximated by this test. The effect of ing those at rupture. From
been observed that, as some overcure in some cases has been recorded at earlier these measurements and the
vulcanized rubber s a m p l e s stages by this test than by the tensile criterion. The cross section of the tongue the
age, their resistance to shear- test is easy to perform and with usual precautions shear-stress-strain curve may
ing and tearing stresses de- should have an accuracy of approximately 10 per cent. be plotted and the relative
creases inuch faster than their energy to start rupture esti-
resistance to tensile stresses, as determined under the stand- mated by determining the area under the curve. Figure 3
ard procedure of rubber-testing. shows the comparison of the shear-stress-strain curves with
During a study of natural and artificial aging some inner- the usual tensile-stress-strain curves. The reasons that the
tube samples, which tested very poorly after the oxygen bomb shear-stress-strain curves do not exactly coincide with the
(50 hours a t 60 C. and 20.4 atm.), were filed for further tensile-stress-strain curves seem to be: (1) Different widths
observation. As these tubes aged the tensile tests showed of samples were stretched; (2) different rates of stretching
less deterioration than was expected. However, a close were employed; (3) near break certain stocks seem to yield
examination of these tubes showed that they had developed some on shearing. In other words, the rupture is very slow
a very poor resistance to tear and were weak when sudden while in other cases it is instantaneous. These curves are
tensile stress was applied. For sake of brevity this lack of compared more fully in Table 111.
resistance t o sudden stress will be called shortness. A shorter and slightly less accurate method of getting the
After several attempts to measure this shortness property relative energy t o rupture the tongue is to take one-half the
without resorting to some new testing machine, the tongue product, S X E/100, where S is the stress a t rupture (kg. per
shear test was developed. It is so named because of the sq. cm.) and E is the per cent ultimate elongation. The
shape of the test specimen and the effect that is produced. product, the shear product, is close enough for ordinary
It seems to give about the proper rating to these short comparisons and closer to the real energy values than the
tubes. tensile product is for the relation that it expresses, be-
cause lower concavity factors exist in the shear-stress-strain
The Test curves.
A specimen like that in Figure 1 is prepared by means of Mechanism of Test
a cutting die, as shown by the pattern in Figure 2. Two
parallel marks exactly 1 inch (2.54 cm.) apart are placed on The manner of rupture and factors influencing the values
the test piece so that they will be near the middle of the obtained were studied to learn, first, what was taking place
parallel section of its tongue. The sample is then placed in as the rubber was strained in this sort of specimen, and second,
1 Presented before the Division of Rubber Chemistry at the 76th
if the test piece was properly proportioned. The rupture
Meeting of the American Chemical Society, Swampscott, Mass , September always takes place a t the end of the tongue (C-D, Figure 1).
10 to 14, 1928. A much smaller expenditure of energy is required to start
18 ANALYTICAL EDI TI0N Vol. 1, No. 1
I I I I
i
Shearing Ultimate Shear
Expt. Thickness IPlies stress eb;y product
'Ii
1.93 0.076 2 15 218 320 25 349
2.94 0.116 3 23 332 410 48 680
E 4.08 0.161 4 24 339 400 48 678
B 2.72 0.107 3 21 303 400 43 606
1.83 0.072 2 22 313 360 39 556
SHEARTESTS
Expt. 1
I
Thickness 1
I
Shearing stress
Ultimate
elonga-
tion
Shear resilience
Kp.-cm. In.-lbs.
% cc. in.8
0.76 0.0300 10.5 152 350 18 257
0.71 0.028b 9.3 132 325 11 159
1.40 0.055 13.5 193 375 26 364
3.12 0.123 18.5 274 440 41 579
24.5 ~
TENSILE
TESTS
Figure 2-Pattern of Ultimate
Figure 1-Test Specimen C u t t i n g Die Expt. Thickness Tensile elonga- Tensile resilience
tion
rupture than would be the case if the same section were Kp.-cm. *.I
broken with the standard dumbbell test specimen. % cc. in.8
0.76 0.030a 114 1575 740 272 3870
Since rubber is so elastic, the mathematical analysis of the 0.71 0.0286 98 1400 740 226 3220
forces that follow the straining of the tongued specimen is 1.40 0.055 136 1900 695 306 4350
3.12 0.123 155 2200 720 418 5950
difficult. Therefore, the specimen shown in Figure 1 was
ruled off into squares before it was put under strain in order a Outside, next to mold face.
b Inside, next to a.
to present a picture of the lines of strain. As the tongue Experiment A-Inner tubes of different thickness. Cured in same heat
(open steam).
stretches out, the legs A and B twist upwards and the dips in Experiment B-Four-ply tube buffed down.
the lines a t C and D become sharper. It is evident that a t Experiment C-Four plies of inner tube stock molded in one slab but
separated with vellum t o give thickness shown. Cure, 75 minutes at 141' C.
these points the rubber suffers the greatest change in direction
of strain and also that a marked shearing action is set up. Various articles2Jr4 have shown how much tensile results
If the cuts a t C and D are rounded with a punch, ruptures are are influenced by the width of the specimen. I n the case of
not obtained a t such low energy values and the test behaves either ring or straight test pieces a decrease of width pro-
more like a tensile test. duces higher tensile values, especially on rich gum stocks.
Studies of the effect of the width of the tongue and of the The Bureau of Standards3 shows that the more even dis-
margin or legs outside the tongue are represented in Figures tribution of fiber stresses in the narrow samples explains the
4 and 5. These curves show that the specimen is so pro- results.
portioned that it avoids the tensile effect as much as possible. The effect of thickness is difficult to determine, because
The narrowing of T with X constant (Figure 4) does not
2 Report of A. C. S. Physical Testing Committee, IND.ENQ.CHEM.,
produce the same type of curve as the widening of X with T 17, 535 (1925).
constant (Figure 5 ) . If the distribution of fiber stresses in 8 Bur. Standards, Bull. 88.
the tongue is taken into account, this difference for tongue 4 Memmler and Schob, International Critical Tables, Vol. 11, p. 266;
widths under 0.5 inch seems to be satisfactorily explained. Mitt. Matrrialfirufungsamt Berlin-Dahlem, 29, 185 (1911).
Figure 3-Stress-Strain
70ELONGATION
Curves of T o n g u e Shear a n d Standard
I TONGUE Wm
l H dT ( X CONSTANT)
TONGUE
SHEAR STANDARD
TENSILE PARBASIS
COMPARISON
SAMPLE Ultimate Ultimate ~
Stress elonga- Shear product Stress elonga- Tensile product Shear Tensile
tion tion product product
-
Kg.
cm.2
-
Lbs.
in.% %
Kg.-cm. -
- cc.
In.-lbs.
in.8
-
Kg.
cm.2
-
Lbs.
in.2 %
Kg.-cm.
cc.
In.-lbs.
in.
Carbon-black tread stock 103 1460 390 200 2847 237 3370 630 1500 21,231 100 100
Carbon-black tap sole 75 1067 280 105 1497 134 1900 335 447 6,365 53 30
Floating red tube, sp. gr. 0.98 36 505 470 84 1186 201 2850 785 1570 22,372 42 105
Molded tube, sp. gr. 1.28 48 682 350 84 1190
Ordinary red tube, sp. gr. 1.06
Truck tube (ZnO), sp. gr. 1.44
24
42
335
590
440
240
52
50
737
708
274
179
239
3900
2650
3400
650
785
578
1840
1410
1380
;:$?!
19,650
42
26
25
123
94
92
Cheap tube (whiting), sp. gr. 1.23 15 215 180 13.7 194 155 2200 596 924 13,112 7 62
Black heel 34 480 130 22 312 94 1340 417 393 5,588 11 26
White suede leather 202 2870 40 40 574 322 4583 50 160 2290 20 11
Manila cardboard 190 2690 0.0 ... .. 197 2800 0.0 .. .... 0 0
such factors as grain effect, heat lag, or heat developed while era1 way with the tearing resistance of such stocks. In gen-
buffing the specimens may introduce serious variations. eral, the effect of pigments on tear resistance is fairly well
Table I shows some data obtained by three different experi- defined by the shear-product value, judging by hand-tear de-
ments. These results leave much to be desired, but indicate terminations. It has also been noted that certain accelerators
that, even in the case of tensile testing of vulcanized rubber, give better results than others.
strictly comparable results are not assured unless the speci-
mens are of nearly the same thickness. Therefore, in per- Tahle 111-Comparison of a Worn a n d a New Tube
YORN NEW
CRITERION EXPLANATION UNIT %BE T F G R~~~~
A A/B
%
Tensile A. C. S. standard Kg./cm.a 150 17c 85
I,bs./in.l 2130 2500 85
Resiiient B y Sheppards inte- Kg.-cm./cc. 213 327
energy gration formula In.-lbs./in.a 3024 4650 67
Tongue shear KgJcm.2 13 18.5
stress (S) I,bs./in.2 186 264 70
Tongue shear Ultimate elongation
strain % (E) 190 380 50
Tongue shear From area under Kg.-cm
In.-lbs.$n.a
cc. 11 34
resilience I shear-strain curve 162 477 34
I-
Tongue shear 1 / a shear stress X
product strain /a S X E Kg.-cm cc. 12 35
100 Or -100 In.-lbs.in.a 176 502 35
//O
Application of Test
The tongue shear test is applicable to prac- 30
tically every type of soft-rubber stock produced.
It suggests many interesting studies, but its
70
most striking applications are: (a) the demon-
stration of the shortness property of aged inner 60 60
tubes, ( B ) its use as a more sensitive index of the SO CTO
extent of artificial aging, and (c) its use with
40 40
other criteria of cure or by itself to detect the
beginning of overcure. Except for showing the JO
% % % %
tensiles. On s e r v i c e
I I
6 df ,le
5LA0 CURES e 141 t
, ,?,
+2119 25
70
54
93
63
96
tests tubes made from
Figure 8-Shear Product us. Tensile or
15 43 51 68
this stock developed Tensile Product on Overcures
17 44 74 89 very poor resistance
4 49 71 76
16 45 77 76 to sudden stress, weak shear resistance, and exceedingly poor
39 42 57 60 tearing qualities.
28 60 78 75
+2 24 23 11 Further studies of its applications are being made.
33 27 80 65
2 50 33 73 Acknowledgment
11 69 66 78
9 40 29 70
50 43 64 64 The writer wishes to express his appreciation to W. E.
Average I 16 45 1 61 69 Glancy, N. E. Tousley, and others who have assisted so much
Size 1, 29 X 4.40; size 2, 31 X 5.25; size 3, 33 X 6.00; size 4, 35 X 5. in the perfection of this test through their helpful suggestions.
HERMAN and others3 have reported one fatality and ide per 10,000 of air is the maximuin concentration to which
% % % %
tensiles. On s e r v i c e
I I
6 df ,le
5LA0 CURES e 141 t
, ,?,
+2119 25
70
54
93
63
96
tests tubes made from
Figure 8-Shear Product us. Tensile or
15 43 51 68
this stock developed Tensile Product on Overcures
17 44 74 89 very poor resistance
4 49 71 76
16 45 77 76 to sudden stress, weak shear resistance, and exceedingly poor
39 42 57 60 tearing qualities.
28 60 78 75
+2 24 23 11 Further studies of its applications are being made.
33 27 80 65
2 50 33 73 Acknowledgment
11 69 66 78
9 40 29 70
50 43 64 64 The writer wishes to express his appreciation to W. E.
Average I 16 45 1 61 69 Glancy, N. E. Tousley, and others who have assisted so much
Size 1, 29 X 4.40; size 2, 31 X 5.25; size 3, 33 X 6.00; size 4, 35 X 5. in the perfection of this test through their helpful suggestions.
HERMAN and others3 have reported one fatality and ide per 10,000 of air is the maximuin concentration to which
potassium iodide solution. Preliminary experiments demon- in the laboratory air, and to determine the effect of tempera-
strated conclusively that it would not be possible satisfac- ture on the evolution of iodine by our iodine pentoxide, the
torily to absorb ethylene at room temperature in a reasonably following experiments were conducted.
short series of gas washing bottles, so a modified gas pipet Air drawn from outside the laboratory and scrubbed
was used for the purpose. The residual gas was diluted with with acid permanganate and sodium hydroxide solutions
air and passed through scrubbers to insure the removal of was passed ovkr iodine pentoxide a t 160" C. to free it from any
the last traces of ethylene. carbon monoxide it might contain. Any iodine evolved was
In order to gain familiarity with the method, several de- then removed from the air by scrubbing with potassium
terminations of carbon monoxide in carbon monoxide-air hydroxide solution, and the resultant moist air was passed
mixtures were carried out. These experiments revealed to over stick potassium hydroxide and then through a drying
us that several important precautions which should be taken tube containing phosphorus pentoxide. From here the air
in the determination have either been overlooked or have passed through a second tube containing iodine pentoxide that
not been sufficiently emphasized. could be kept a t any desired temperature. Iodine evolved was
Preparation of Iodine Pentoxide
trapped in 5 cc. of potassium iodide containing 1 cc. of 1 per
cent soluble starch, and t,itrated a t the conclusion of the ex-
The purity of the iodine pentoxide is probably the most periment with 0.0001 N sodium arsenite. In each experiment
-
imuortaiit single factor necessary to success. We wish to 2.3 liters of dry carbon monoxide-free air mere Dassed over the
emphasize again, as have iodine pentoxide. The re-
D a v i e s and Hartley6 and sults of the experiments are
Edell,' that iodine pentoxide A method of determining small amounts of carbon expressed in Figure I, which
to be used for the determina- monoxide in ethylene has been developed. Ethylene is shows the amount of iodine
tion of small amounts of first removed from the gas under examination by means found in t h e p o t a s s i u m
carbon monoxide should be of 25 per cent oleum, and the residual gas, diluted with iodide solution plotted, as
p r e p a r e d from iodic acid air, is freed successively from traces of ethylene by 60 cubic centimeters of 0.001 N
made by chloric acid oxida- per cent oleum, from SOa by 98 per cent sulfuric acid, iodine, against temperature.
tion of p u r e r e s u b l i m e d from SOz by soda lime, and from moisture by Pz05,and It is significant to note that
iodine.8 Too m u c h c a r e is finally passed over hot 1205, with subsequent titra- the magnitude of the blank
cannot be taken in prepar- tion of the iodine liberated. The more important pre- obtained in a carbon mon-
ing the reagent, and pent- cautions in the determination of carbon monoxide in oxide determination will be
oxide ~ d i i c hhas been care- air by means of the iodine pentoxide method are dis- affected little by tempera-
lessly p r e p a r e d mill give cussed. The method has been used for determining ture, provided the analysis
blanks that v a r y widely carbon monoxide in mixtures containing as little as 5 is conducted at a tempera-
from day to day and are parts of carbon monoxide per 100,000 of mixture, and t u r e below 180" C. and
much higher than those ob- could be satisfactorily employed with even lower con- above the temperature a t
tained with a good sample centrations. which o x i d a t i o n of t h e
of reagent. Lamb's method monoxide is complete. We
of pieparing iodine pent- chose 170" C. as our work-
oside ifas used in making ing temperature be cause
our reagent, and was founud t o be the only method on which there was available a simple means of keeping the oil bath
me could depend to give a uniformly satisfactory product. a t that temperature. There is no reason t o believe that any
other temperature in this neighborhood would not be equally
Effect of Temperature on Decomposition of Iodine satisfactory. It would not, however, be advisable to work
Pentoxide above 180" C., as the rate of decomposition of iodine pent-
oxide can be seen from the figure to increase rapidly above
The temperature a t which iodine pentoxide will completely 180" C. From available work and our own experience it is
oxidize carbon monoxide was determined by Gautier t o be evident that oxidation of carbon monoxide to carbon dioxide
100" C.9 Various investigators since that time have em- is rapid and complete a t 170" C.
ployed temperatures varying from 100" t o 200" C.lo No
one, however, has taken the trouble t o determine systemati- Drying the Gas Sample
cally the manner in which temperature affects the decom-
position of iodine pentoxide. Obviously this decomposition The necessity of having dry gas has not been emphasized
of the reagent is an important factor in the determination, sufficiently. Better results in our analyses were always
because it must be uniform a t any given temperature if obtained when gas dried over phosphorus pentoxide was
dependable results are to be obtained. Furthermore, it used. The single point in Figure 1 labeled "moist air"
would be very desirable t o have the amount of iodine liberated was obtained in the same way as the points of the other curve,
by decomposition of the reagent a relatively small fraction of except that air humidified by bubbling through 10 per cent
the total iodine evolved in a determination. sodium hydroxide solution was passed directly over iodine
Our experience is in accord with that of Edell,' who found pentoxide a t 170" C. without drying.
that the best iodine pentoxide he could make liberated small In our first experiments 98 per cent sulfuric acid was used
quant,ities of iodine on heating. I n order t o determine t o dry the gas. In one experiment, however, enough iodine
whether, as Seidell has claimed,'l this evolution of iodine is mas added to the absorption bulb a t the beginning of the
caused by the presence of small amounts of carbon monoxide run to give a color with the starch. As the experiment pro-
ceeded the color of the indicator gradually bleached, and
6 Davies a n d Hartley, J . SOC.Chem. Ind., 45, 164 (1926).
actually several drops of 0.001 iV iodine were needed to restore
7 Edell, IND. ENG.CHEM.,20, 275 (1928).
Lamb e t al., J . A m . Chem. Soc., 42, 1636 (1920).
the blue color a t the conclusion of the test. It was necessary,
8 Compt. rend., 126, 793, 931, 1299 (1898). therefore, to introduce in the scrubbing train, subsequent to
10 Tausz and Jungmann (Gas Wessevfech, 70, 1049 (1927)) recommend a the 98 per cent sulfuric acid, soda lime and then phosphorus
temperature of 120-130 C. if the gas contains hydrogen, and one of 195' C. pentoxide. Thorburn12 has also noted that concentrated
if no hydrogen is present.
ENG.CHEM.,6, 321 (1914).
11 Seidell, J. IND. 12 Thorburn, J . SOC.Chem. Ind., 46, 355 (1927)
22 ANALYTICAL EDITION Vol. 1, No. 1
sulfuric acid is unsatisfactory as a final drying agent in this absorbed in boiled potassium hydroxide solution (2:3) and
determination. about 100 cc. of monoxide collected in a gas buret and aE
Titration of the Iodine lowed to stand over dilute potassium hydroxide for several
days. Measurement of the residual gas after shaking a
The use of 0.001 N sodium thiosulfate solutions was soon known amount of the carbon monoxide with ammoniacal
abandoned. Even when made from carefully boiled distilled cuprous chloride showed the purity of the gas. As the im-
water, thiosulfate solutions decomposed far more rapidly on purity was undoubtedly air, the volume of residual gas was
standing than did arsenite solutions of similar strength. multiplied by five-fourths in calculating the purity of the
Figure 2 gives a good idea of the relative stability of the two original sample, as oxygen is absorbed by the ammoniacal
solutions. Although the 0.001 N arsenite solutions did not cuprous chloride.
decompose appreciably up to the end of 5 days, we chose to
Apparatus and Method
make up our 0.001 N and 0.0001 N solutions fresh each day
from a 0.1 N solution whose normality was checked at fre- The apparatus used is shown in Figure 3. The measuring
quent intervals throughout the duration of the experiments. bulb (7), carefully calibrated and water-jacketed, had a
volume of 526.25 cc. a t 22.1" C., and contained mercury as
a sealing fluid. It was connected to the leveling bulb (5).
For removing ethylene, an absorption pipet (6), containing
glass rods and filled with fuming sulfuric acid (25 per cent
SO3),was provided. The mercury trap (8), with its attached
manometer (9), and the U tube (10) containing 98 per cent
sulfuric acid served to prevent any fumes from the 60 per
cent oleum in the scrubbing train (11, 12, 13) from getting
into the measuring bulb. A trap (14) containing 98 per cent
sulfuric acid served to protect the iodine pentoxide from
oleum fumes, and the drying tube (15), containing soda lime
in one arm and phosphorus pentoxide in the other, prevented
the entrance of sulfur dioxide from the sulfuric acid and mois-
ture into the iodine pentoxide tube. The iodine pentoxide
was contained in a 5-mm. glass tube coiled as shown in the
figure (16) to allow preheating the gas. The tube contained
iodine pentoxide for about 5 inches of its length, in particles
of such a size that they would pass through a 15-mesh sieve
Figure 1-Effect of Temperature on Stability of and be retained by a 20-mesh sieve. An electric hot plate
Iodine Pentoxide
maintained the Crisco bath a t 170" C. The exit from the
Sodium arsenite solutions which were 0.001 N a n d 0.0001 N iodine pentoxide tube was drawn down to small bore and
were used interchangeably. It was found more convenient inserted into the inlet of the absorption apparatus (17),
to reach the proper end point with the more dilute solution, which contained potassium iodide starch solution. The
and 0.0001 N arsenite was used in most of the work. Numer-
ous experiments with 0.001 N and 0.0001 N solutions showed
that no appreciable error was introduced by the use of the
more dilute solution. The titration was carried out in the
absorption apparatus to avoid the necessity of dilution during
transfer to another container.
Purity of Ethylene
The ethylene used in the investigation was obtained from a
cylinder of compressed gas. Blank runs using one bulb of
ethylene followed by two of air showed iodine evolved equiva-
lent to 1.3 cc. of 0.0001 N arsenite. Although this blank
value is larger than that obtained when three bulbs of air
were used,l3 if we calculate from the iodine evolved the
amount of carbon monoxide present in the ethylene, this
amount can be no greater than 1 part per 100,000 of ethylene. Figure 2-Relative Stability of 0.001 N S o d i u m
Furthermore, i t may be that some impurity other than carbon Arsenite and Thiosulfate Solutions
monoxide in the gas caused this greater evolution of iodine.
I n a future communication we will give the results of de- joint was then covered with pressure tubing. There were
terminations run on ethylene which has been purified by no rubber connections from the measuring bulb to the joint
fractional distillation. between the iodine pentoxide tube and absorbing apparatus.
The lower limit of certain detection of carbon monoxide in All stopcocks were lubricated with a little phosphorus pentox-
ethylene by the official hemoglobin test is 0.02 per cent. ide moistened with water. The absorbing solution was 5 cc.
The ethylene used here, therefore, contained no more than of a 1 per cent potassium iodide solution and 1 cc. of a 1 per
0.05 of this amount. As was to be expected, it gave a nega- cent solution of soluble starch, both freshly prepared.
tive test by the official hemoglobin method. I n order to familiarize ourselves with the operation of the
apparatus, several determinations of carbon monoxide in
Preparation and Purity of Carbon Monoxide carbon monoxide-air mixtures were made. As we believe
Carbon monoxide was prepared by the action of concen- that the results are more accurate than most of those in the
trated sulfuric acid on oxalic acid. The carbon dioxide was literature, particularly in view of the small size of the sample
18 Iodine was evolved in such determinations equivalent to 0.55 cc. of
used, they are included with those obtained from carbon
0,0001 N arsenite. monoxide-ethylene mixtures.
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 23
OSTROL analysis in the annual production of over T o the filtrate from magnesium hydroxide, sodium carbon-
and the cover are washed, 4 drops of 0.1 per cent alcohol 300 mg. of CaO, the magnesium was precipitated as the hy-
solution of dimethylaminobenzene are added, and the solu- droxide and the precipitates were filtered off with lighted Bunsen
burners at full height on each side 10 cm. from the funnels.
tion is titrated with 0.25 N sodium hydroxide to a slight pink In one case the beaker and funnel containing the magnesium
color of the indicator. The number of cubic centimeters of hydroxide were left uncovered and the beaker was allowed to
'acid required to dissolve the precipitate of calcium carbonate stand by one burner. In the other case both beaker and funnel
multiplied by 1.4 gives the per cent of CaO in the sample. were kept covered except while the funnel was being filled. The
magnesium was determined as described above and the calcium
Table I-Representative Analyses of Cementa present with the magnesium hydroxide precipitate was deter-
RZOS MAGNESIUM
OXIDE CALCIUM OXIDE mined as the oxalate.
SAMPLEPresent Found Present Found Present Found
1
Per cent
9.3
Per cent
9.4
Per cent
4.9 5.0
Per cent
61.8
Per cent Per cent
62.0
The values obtained are given below:
2 8.9 8.9 1.1 1.1 64.0 64.5
3 10.5 10.4 3.1 3.3 61.7 61.9 UNCOVERED COVERED
4 9.2 9.3 1.5 1.5 62.5 82.7 SAMPLE SAMPLE
a Acknowledgment is made to the Bureau of Standards, from whom the 1Mg. Mg.
analyzed samples were obtained. Magnesium 22 8 21.3
Magnesium equivalent of calcium 33 2.1
Discussion of Method
It is quite evident from these results that precautions should
I n the calculation of results, for strict accuracy, the num- be taken to prevent carbonation. However, if due precau-
ber of cubic centimeters of acid required to dissolve the mag- tions are taken, no great error should be introduced, for even
nesium hydroxide should be multiplied by 1.008 instead of with the excellent chance for carbonation in the covered sam-
1, and the number of cubic centimeters of acid required to ple, the magnesium found was only 1.3 mg. higher than the
dissolve the calcium carbonate should be multiplied by 1.4018 amount present. I n a 0.5-gram sample of cement, this corre-
instead of 1.4. However, since this method will be of use sponds to a little over 0.4 per cent of MgO.
only for control work, where the highest accuracy is not re- The determination of iron and aluminum oxides by this
quired, the approximations given will be satisfactory. method takes a little longer than by two precipitations with
The fact that the results by this method are as near cor- ammonium hydroxide, for when this method is used the
rect as shown in the above representative set of analyses is first washing must be thorough enough to insure practically
due to a balancing of errors. The loss of magnesium hydrox- complete removal of calcium and magnesium. This is be-
ide in the filtrate is offset by the precipitate of calcium car- cause the filtrate from the second precipitation contains
bonate, due to the slight amount of carbon dioxide absorbed ammonium hydroxide, which, if mixed with the original
from the air, and to the presence of a small amount of sodium filtrate, will interfere with subsequent tests. The accuracy
hydroxide not washed out of the precipitated magnesium of the two methods should be about the same.
hydroxide. After magnesium was determined, the solu- This method for the determination of calcium and mag-
tions were filtered and the calcium determined by precipi- nesium is much more rapid than the customary procedure
tation of the oxalate and titration with potassium perman- of precipitating the calcium as the oxalate and the magnesium
ganate. The results, expressed in terms of MgO, were 0.3, as magnesium-ammonium phosphate, particularly if a double
0.3, 0.4, and 0.2 per cent, respectively, for samples 1, 2, 3, precipitation of magnesium is carried out. I n this procedure
and 4. The slightly high results for calcium, in spite of this no evaporation is necessary, and both calcium and magnesium
loss, must be due to the sodium carbonate remaining after are determined volumetrically with one set of standard solu-
three washings. tions. However, it is not quite so accurate as the customary
I n order to cut carbonation oi the calcium to a minimum, procedure, for calcium oxalate can be washed more thoroughly
the precipitate of magnesium hydroxide should be filtered than can calcium carbonate, without danger of loss, and in
in a room in which there are few burners, and the beaker spite of balancing errors, the results for magnesium are not
and funnel containing the precipitate should be kept covered so accurate as those obtained by the more tedious gravimetric
as much as possible. To see what would be the effect of method.
filtering in the presence of an abundance of carbon dioxide, This procedure should be of especial advantage for control
the following experiment was carried out: work, where magnesium alone is determined. When ex-
In two solutions, each containing 10 cc. of magnesium chloride treme accuracy is not required, it should find a field of useful-
solution (1 cc. = 2 mg. Mg) and calcium chloride equivalent to ness, also, when Rz03and calcium are to be determined.
-RAY methods have been applied successfullyto quan- gasolinesvaries from under 0.5 to 3 cc. per gallon (3.7854 liters),
ethyl lead in gasoline was suggested to the writers by George creased to the accuracy required in commercial analysis.
Calingaert of the Ethyl Gasoline Corporation. 0.1 cc. per gallon is equivalent to 1 part in 14,000 parts by
The concentration range of tetraethyl lead in commercial weight. Such high sensitivity is possible only when the con-
1 Received August 20, 1928. stituents (in this case solvent and solute) differ very widely
26 ANALYTICAL EDITION Vol. 1, No. 1
and the cover are washed, 4 drops of 0.1 per cent alcohol 300 mg. of CaO, the magnesium was precipitated as the hy-
solution of dimethylaminobenzene are added, and the solu- droxide and the precipitates were filtered off with lighted Bunsen
burners at full height on each side 10 cm. from the funnels.
tion is titrated with 0.25 N sodium hydroxide to a slight pink In one case the beaker and funnel containing the magnesium
color of the indicator. The number of cubic centimeters of hydroxide were left uncovered and the beaker was allowed to
'acid required to dissolve the precipitate of calcium carbonate stand by one burner. In the other case both beaker and funnel
multiplied by 1.4 gives the per cent of CaO in the sample. were kept covered except while the funnel was being filled. The
magnesium was determined as described above and the calcium
Table I-Representative Analyses of Cementa present with the magnesium hydroxide precipitate was deter-
RZOS MAGNESIUM
OXIDE OXIDE
CALCIUM mined as the oxalate.
SAMPLEPresent Found Present Found Present Found
Per cent Per cent Per cent Per cent Per cent Per cent The values obtained are given below:
1 9.3 9.4 4.9 5.0 61.8 62.0
2 8.9 8.9 1.1 1.1 64.0 64.5
3 10.5 10.4 3.1 3.3 61.7 61.9 UNCOVERED COVERED
4 9.2 9.3 1.5 1.5 62.5 82.7 SAMPLE SAMPLE
a Acknowledgment is made to the Bureau of Standards, from whom the 1Mg. Mg.
analyzed samples were obtained. Magnesium 22 8 21.3
Magnesium equivalent of calcium 33 2.1
Discussion of Method
It is quite evident from these results that precautions should
I n the calculation of results, for strict accuracy, the num- be taken to prevent carbonation. However, if due precau-
ber of cubic centimeters of acid required to dissolve the mag- tions are taken, no great error should be introduced, for even
nesium hydroxide should be multiplied by 1.008 instead of with the excellent chance for carbonation in the covered sam-
1, and the number of cubic centimeters of acid required to ple, the magnesium found was only 1.3 mg. higher than the
dissolve the calcium carbonate should be multiplied by 1.4018 amount present. I n a 0.5-gram sample of cement, this corre-
instead of 1.4. However, since this method will be of use sponds to a little over 0.4 per cent of MgO.
only for control work, where the highest accuracy is not re- The determination of iron and aluminum oxides by this
quired, the approximations given will be satisfactory. method takes a little longer than by two precipitations with
The fact that the results by this method are as near cor- ammonium hydroxide, for when this method is used the
rect as shown in the above representative set of analyses is first washing must be thorough enough to insure practically
due to a balancing of errors. The loss of magnesium hydrox- complete removal of calcium and magnesium. This is be-
ide in the filtrate is offset by the precipitate of calcium car- cause the filtrate from the second precipitation contains
bonate, due to the slight amount of carbon dioxide absorbed ammonium hydroxide, which, if mixed with the original
from the air, and to the presence of a small amount of sodium filtrate, will interfere with subsequent tests. The accuracy
hydroxide not washed out of the precipitated magnesium of the two methods should be about the same.
hydroxide. After magnesium was determined, the solu- This method for the determination of calcium and mag-
tions were filtered and the calcium determined by precipi- nesium is much more rapid than the customary procedure
tation of the oxalate and titration with potassium perman- of precipitating the calcium as the oxalate and the magnesium
ganate. The results, expressed in terms of MgO, were 0.3, as magnesium-ammonium phosphate, particularly if a double
0.3, 0.4, and 0.2 per cent, respectively, for samples 1, 2, 3, precipitation of magnesium is carried out. I n this procedure
and 4. The slightly high results for calcium, in spite of this no evaporation is necessary, and both calcium and magnesium
loss, must be due to the sodium carbonate remaining after are determined volumetrically with one set of standard solu-
three washings. tions. However, it is not quite so accurate as the customary
I n order to cut carbonation oi the calcium to a minimum, procedure, for calcium oxalate can be washed more thoroughly
the precipitate of magnesium hydroxide should be filtered than can calcium carbonate, without danger of loss, and in
in a room in which there are few burners, and the beaker spite of balancing errors, the results for magnesium are not
and funnel containing the precipitate should be kept covered so accurate as those obtained by the more tedious gravimetric
as much as possible. To see what would be the effect of method.
filtering in the presence of an abundance of carbon dioxide, This procedure should be of especial advantage for control
the following experiment was carried out: work, where magnesium alone is determined. When ex-
In two solutions, each containing 10 cc. of magnesium chloride treme accuracy is not required, it should find a field of useful-
solution (1 cc. = 2 mg. Mg) and calcium chloride equivalent to ness, also, when Rz03and calcium are to be determined.
-RAY methods have been applied successfullyto quan- gasolinesvaries from under 0.5 to 3 cc. per gallon (3.7854 liters),
ethyl lead in gasoline was suggested to the writers by George creased to the accuracy required in commercial analysis.
Calingaert of the Ethyl Gasoline Corporation. 0.1 cc. per gallon is equivalent to 1 part in 14,000 parts by
The concentration range of tetraethyl lead in commercial weight. Such high sensitivity is possible only when the con-
1 Received August 20, 1928. stituents (in this case solvent and solute) differ very widely
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 27
a
X-RAY TUBE
thickness of specimen penetrated by the beam. Figure 2
I /LEAD
PROTECTIVE RESISTANCE
shows two curves obtained under the above operating condi-
SAMPLE C A LVAN OMETER tions. From these curves i t seems evident that: (1) The
CONTAINER ., SHUNT method holds distinct promise as an analytical tool; (2) the
SULPHUR
experimental conditions as outlined give practically straight-
INSULATOR line curves; (3) an accurate control of operating conditions
- L---J is important. It was also observed that close current control
H.V.
IONIZATION
is equally important,
CHAMBER Among further re-
finements to increase
-
- !
BATTERY OF the precision may be
100 VOLTS
mentioned a high
Figure 1
sensitivity resistance
A diagram of the experimental arrangement is shown in bridge in conjunction
Figure 1. All wires, batteries, and instruments were electri- with an i o n i z a t i o n
cally shielded. A water-cooled molybdenum target diffrac- chamber, or two ioni- 4
tion type Coolidge tube, operating a t 35 kilovolts and 20 zation chambers bal- 2
milliamperes, was found well suited for the work. The ioni- a n c e d a g a i n s t each
zation chamber was of the Bragg type and the galvanometer other, one for a stand- ti
was a Leeds and Northrup moving-coil type instrument with ard specimen and the a
I O 20 IS2 30
a sensitivity of 10-10 amperes. Specimen containers made other for t h e un- ccCONC Pb(ET)*PEI CAL
from brass tubes having an inside diameter of 3.97 cm. with known. Figure 2
an effective length of 5.08 em. were found satisfactory. The As the time required
capped ends contained 2.54 cm. diameter holes, over which for each determination by this method may reasonably be an-
0.00508 cm. thick aluminum windows were placed. ticipated t o be as short as 5 minutes, while the usual chemi-
The standard solutions were made up from tetraethyl cal method requires 40 minutes, a further practical and
fluid of such a composition that 3 cc. of fluid contained economic point is added to the value of this new method.
a
X-RAY TUBE
thickness of specimen penetrated by the beam. Figure 2
I /LEAD
PROTECTIVE RESISTANCE
shows two curves obtained under the above operating condi-
SAMPLE C A LVAN OMETER tions. From these curves i t seems evident that: (1) The
CONTAINER ., SHUNT method holds distinct promise as an analytical tool; (2) the
SULPHUR
experimental conditions as outlined give practically straight-
INSULATOR line curves; (3) an accurate control of operating conditions
- L---J is important. It was also observed that close current control
H.V.
IONIZATION
is equally important,
CHAMBER Among further re-
finements to increase
-
- !
BATTERY OF the precision may be
100 VOLTS
mentioned a high
Figure 1
sensitivity resistance
A diagram of the experimental arrangement is shown in bridge in conjunction
Figure 1. All wires, batteries, and instruments were electri- with an i o n i z a t i o n
cally shielded. A water-cooled molybdenum target diffrac- chamber, or two ioni- 4
tion type Coolidge tube, operating a t 35 kilovolts and 20 zation chambers bal- 2
milliamperes, was found well suited for the work. The ioni- a n c e d a g a i n s t each
zation chamber was of the Bragg type and the galvanometer other, one for a stand- ti
was a Leeds and Northrup moving-coil type instrument with ard specimen and the a
I O 20 IS2 30
a sensitivity of 10-10 amperes. Specimen containers made other for t h e un- ccCONC Pb(ET)*PEI CAL
from brass tubes having an inside diameter of 3.97 cm. with known. Figure 2
an effective length of 5.08 em. were found satisfactory. The As the time required
capped ends contained 2.54 cm. diameter holes, over which for each determination by this method may reasonably be an-
0.00508 cm. thick aluminum windows were placed. ticipated t o be as short as 5 minutes, while the usual chemi-
The standard solutions were made up from tetraethyl cal method requires 40 minutes, a further practical and
fluid of such a composition that 3 cc. of fluid contained economic point is added to the value of this new method.
show distinct gradations in color. With these larger quan- and standard. It is necessary to evaporate the whole amount
tities of BOs it may be advisable to use 0.5 cc. of turmeric solu- of each down a t once, rather than to evaporate a larger
tion instead of 1 cc., provided, of course, the same quantity is volume of t,he sample down in portions for comparison with
added t o both sample and standard. standards prepared from a volume of 10 cc.
If the sample is high in dissolved mineral matter, the ac- Xatural waters may dissolve appreciable quantities of
cumulation of salts in the bottom of the dish may make diffi- boron from the glass bottles in which they are collected, es-
cult a comparison with standards prepared from distilled pecially if kept in the bottles some time before testing for
water. With these samples small quantities of the salt solu- borax. It is advisable, therefore, t o make the test for borax
tion, the quantity depending upon the mineral content of as soon as possible after collection of the Pamples unless the
the samples, may be added to the standards t o make them bottles are known t o be satisfactory in this respect. Blank
more nearly comparable with the samples. It is difficult, tests for borax in distilled water which has stood for several
however, t o obtain a similar deposition of salts in both sample days in the bottles to be used will indicate whether or not
and standards. If the accumulation of salts is too great for borax is given up by the bottles.
comparison with the ordinary standards, it may be better,to The salt solution used t o make the residue upon evapora-
use a smaller volume of the sample, diluted to 10 cc., for the tion of the standards more nearly comparable with the samples,
determination, rather than to try to duplicate the salt deposi- if the samples are high in dissolved mineral matter, should
tion. also be checked from time to time for borax dissolved from
If the borate content of the sample is very low, it is some- the bottle. A fresh salt solution is preferable t o one that
times advisable to use 25 cc. as the volume for both sample has been standing in the bottle for some time.
S A reagent for the detection and estimation of nitrites lized and then acted upon by pyridine-free ammonium
show distinct gradations in color. With these larger quan- and standard. It is necessary to evaporate the whole amount
tities of BOs it may be advisable to use 0.5 cc. of turmeric solu- of each down a t once, rather than to evaporate a larger
tion instead of 1 cc., provided, of course, the same quantity is volume of t,he sample down in portions for comparison with
added t o both sample and standard. standards prepared from a volume of 10 cc.
If the sample is high in dissolved mineral matter, the ac- Xatural waters may dissolve appreciable quantities of
cumulation of salts in the bottom of the dish may make diffi- boron from the glass bottles in which they are collected, es-
cult a comparison with standards prepared from distilled pecially if kept in the bottles some time before testing for
water. With these samples small quantities of the salt solu- borax. It is advisable, therefore, t o make the test for borax
tion, the quantity depending upon the mineral content of as soon as possible after collection of the Pamples unless the
the samples, may be added to the standards t o make them bottles are known t o be satisfactory in this respect. Blank
more nearly comparable with the samples. It is difficult, tests for borax in distilled water which has stood for several
however, t o obtain a similar deposition of salts in both sample days in the bottles to be used will indicate whether or not
and standards. If the accumulation of salts is too great for borax is given up by the bottles.
comparison with the ordinary standards, it may be better,to The salt solution used t o make the residue upon evapora-
use a smaller volume of the sample, diluted to 10 cc., for the tion of the standards more nearly comparable with the samples,
determination, rather than to try to duplicate the salt deposi- if the samples are high in dissolved mineral matter, should
tion. also be checked from time to time for borax dissolved from
If the borate content of the sample is very low, it is some- the bottle. A fresh salt solution is preferable t o one that
times advisable to use 25 cc. as the volume for both sample has been standing in the bottle for some time.
S A reagent for the detection and estimation of nitrites lized and then acted upon by pyridine-free ammonium
amino compound excelled the other in several respects. The sensitivity. It is also evident that standards of this reagent
methylated reagent is more sensitive and far more stable, are permanent, thus obviating the necessity for preparing
and its action in producing a color change much more rapid. the usual unstable fuchsin standards.
During a period of sixty days the depth of the color was
identical with that originally produced. The contents of Effect of Hydrogen Sulfide
each of the Nessler tubes had been protected from dust dur- A series of experiments was carried out using solutions
ing this interval, and each had remained unchanged in appear- similar in concentration to those described above. Each
ance. Although the tubes were not exposed t o direct sun- contained a small proportion of hydrogen sulfide, in incre-
light during this time, no attempt was made to exclude light. ments of 0.01 p. p. m. The results obtained in these tests
Preparation of Dimethyl-a-hTaphthylamineReagent assured one that while dimethyl-a-naphthylamine was
noticeably affected by large quantities of hydrogen sulfide,
A solution of 6.25 grams of dimethyl-a-naphthylamine the effect was not so apparent nor did it occur with the same
dissolved in 1 liter of 4 N acetic acid in 95 per cent methanol readiness as when a-naphthylamine was used. Indeed, in
insures the most satisfactory results in the colorimetric the lower concentrations of the gas, it was impossible t o detect
estimation of nitrites. Such a solution is stable and of high any color change in the solutions under examination.
guti, and M e y e r h o f f e r . 8
Several of them state that un-
the determination of sulfur in particularly in the determination of sulfur in rubber der certain conditions barium
rubber, and suggests that, in compounds in conjunction with the nitric-perchloric sulfate can be decomposed
the presence of barium, lead, acid method. completely, i n t h i s w a y .
or other c o m p o u n d s w h i c h It has been found that barium sulfate can be de- The required conditions, how-
yield an insoluble residue after composed to the extent of over 99 per cent in a single ever, are hardly suitable for
oxidation of the rubber by the operation, by boiling for 1 hour (or 2 hours in the case the purposes of modern quan-
acid oxidizing agents, this in- of barite) with a strong solution containing at least t i t a t i v e analysis. Further-
soluble residue be filtered off fifteen times the theoretical amount of sodium car- more, since there are some
a n d f u s e d separately with bonate. conflicting statements in the
sodium carbonate. Although this degree of decomposition is not so high literature on this subject, an
I n place of this fusion, how- as that attained by the fusion method, and is not suffi- investigation was undertaken
ever, it was suggested3 that cient for the requirements of the most exacting quanti- to determine how completely
any sulfur that might be pres- tative analysis, it is considered sufficient for many pur- the barium sulfate can be de-
ent in this insoluble residue in poses, particularly in the analysis of materials con- composed by aqueous solu-
the form of sulfates might be taining only small amounts of barium sulfate, besides tions of sodium carbonate un-
extracted with a solution of offering the advantages of speed and simplicity of der conditions which would
sodium carbonate. The idea operation. make the Drocedure suitable
appeared very attractive, for, for ordinary analytical work,
in case this procedure were found feasible, the decomposition of and also to determine the necessary conditions for this con-
insoluble sulfates could be accomplished in the same flask in version.
which the original rubber sample was oxidized, thus materially
simplifying the procedure as well as saving time and labor. Method
No difficulty was anticipated in decomposing the sulfates The method of procedure was, in general, t o decompose a
of lead and calcium in this way, sufficiently for analytical known amount of sodium sulfate with barium chloride, heat
purposes, for both of these salts are more soluble than the corre- the barium sulfate thus obtained with varying amounts of
sponding carbonates; but in the case of barium sulfate the sodium carbonate solution, filter, and determine the a,mount
situatioii is quite different, for this compound is much less of sulfate in the filtrate by t.he usual method of precipitating
soluble than the corresponding carbonate. with barium chloride. Since, however, the writer was par-
It has been known for a long time that barium sulfate can be ticularly interested in the availability of the procedure in con-
decomposed more or less completely, under suitable condi- nection with the nitric-perchloric acid method of determining
tions, by aqueous solutions of alkali carbonates; in fact this sulfur in rubber, the conditions were slightly modified in
reaction has been suggested in various schemes of qualitative order t o approach more nearly those that would be encountered
analysis for the detection of sulfates. The reaction has been in that method. It is believed that these conditions would
studied by a number of investigators a t different times,
4 Dulong, Ann. chim. phys., [ l ] 82, 273 (1812).
1 Received June 15, 1928. Publication approved by the Director of 5 Rose, Ann. Physik (Pogg.), 94, 481 (1855); 96, 96, 284 (1855).
the Bureau of Standards of the U. S. Department of Commerce. 8 Cf. Nernst, Lehrbuch der theoretischen Chemie, 2 Aufl., p. 409
2 IND.ENG. CHEM.,20, 1234 (1928). 7 Malaguti, Ann. chim. phys., [3] 37, 198 (1853); 61, 328 (1857).
8 By A. T. McPherson, of the Bureau of Standards. 8 Meyerboffer, 2. physik. Chem., S3, 513 (1905).
January 15, 1929 I-YDUSTRIAL Ah-D ENGI,VEERING CHEMIXTRY 29
amino compound excelled the other in several respects. The sensitivity. It is also evident that standards of this reagent
methylated reagent is more sensitive and far more stable, are permanent, thus obviating the necessity for preparing
and its action in producing a color change much more rapid. the usual unstable fuchsin standards.
During a period of sixty days the depth of the color was
identical with that originally produced. The contents of Effect of Hydrogen Sulfide
each of the Nessler tubes had been protected from dust dur- A series of experiments was carried out using solutions
ing this interval, and each had remained unchanged in appear- similar in concentration to those described above. Each
ance. Although the tubes were not exposed t o direct sun- contained a small proportion of hydrogen sulfide, in incre-
light during this time, no attempt was made to exclude light. ments of 0.01 p. p. m. The results obtained in these tests
Preparation of Dimethyl-a-hTaphthylamineReagent assured one that while dimethyl-a-naphthylamine was
noticeably affected by large quantities of hydrogen sulfide,
A solution of 6.25 grams of dimethyl-a-naphthylamine the effect was not so apparent nor did it occur with the same
dissolved in 1 liter of 4 N acetic acid in 95 per cent methanol readiness as when a-naphthylamine was used. Indeed, in
insures the most satisfactory results in the colorimetric the lower concentrations of the gas, it was impossible t o detect
estimation of nitrites. Such a solution is stable and of high any color change in the solutions under examination.
N A previous paper2 the A study has been made of the decomposition of among them D ~ l o n gRose,5,~
I author describes an acid
solution method of de-
composing rubber samples in
barium sulfate by aqueous solutions of sodium car-
bonate with the view of the possible utilization of this
reaction in quantitative analysis in place of fusions,
Guldberg and Waage,6 Mala-
guti, and M e y e r h o f f e r . 8
Several of them state that un-
the determination of sulfur in particularly in the determination of sulfur in rubber der certain conditions barium
rubber, and suggests that, in compounds in conjunction with the nitric-perchloric sulfate can be decomposed
the presence of barium, lead, acid method. completely, i n t h i s w a y .
or other c o m p o u n d s w h i c h It has been found that barium sulfate can be de- The required conditions, how-
yield an insoluble residue after composed to the extent of over 99 per cent in a single ever, are hardly suitable for
oxidation of the rubber by the operation, by boiling for 1 hour (or 2 hours in the case the purposes of modern quan-
acid oxidizing agents, this in- of barite) with a strong solution containing at least t i t a t i v e analysis. Further-
soluble residue be filtered off fifteen times the theoretical amount of sodium car- more, since there are some
a n d f u s e d separately with bonate. conflicting statements in the
sodium carbonate. Although this degree of decomposition is not so high literature on this subject, an
I n place of this fusion, how- as that attained by the fusion method, and is not suffi- investigation was undertaken
ever, it was suggested3 that cient for the requirements of the most exacting quanti- to determine how completely
any sulfur that might be pres- tative analysis, it is considered sufficient for many pur- the barium sulfate can be de-
ent in this insoluble residue in poses, particularly in the analysis of materials con- composed by aqueous solu-
the form of sulfates might be taining only small amounts of barium sulfate, besides tions of sodium carbonate un-
extracted with a solution of offering the advantages of speed and simplicity of der conditions which would
sodium carbonate. The idea operation. make the Drocedure suitable
appeared very attractive, for, for ordinary analytical work,
in case this procedure were found feasible, the decomposition of and also to determine the necessary conditions for this con-
insoluble sulfates could be accomplished in the same flask in version.
which the original rubber sample was oxidized, thus materially
simplifying the procedure as well as saving time and labor. Method
No difficulty was anticipated in decomposing the sulfates The method of procedure was, in general, t o decompose a
of lead and calcium in this way, sufficiently for analytical known amount of sodium sulfate with barium chloride, heat
purposes, for both of these salts are more soluble than the corre- the barium sulfate thus obtained with varying amounts of
sponding carbonates; but in the case of barium sulfate the sodium carbonate solution, filter, and determine the a,mount
situatioii is quite different, for this compound is much less of sulfate in the filtrate by t.he usual method of precipitating
soluble than the corresponding carbonate. with barium chloride. Since, however, the writer was par-
It has been known for a long time that barium sulfate can be ticularly interested in the availability of the procedure in con-
decomposed more or less completely, under suitable condi- nection with the nitric-perchloric acid method of determining
tions, by aqueous solutions of alkali carbonates; in fact this sulfur in rubber, the conditions were slightly modified in
reaction has been suggested in various schemes of qualitative order t o approach more nearly those that would be encountered
analysis for the detection of sulfates. The reaction has been in that method. It is believed that these conditions would
studied by a number of investigators a t different times,
4 Dulong, Ann. chim. phys., [ l ] 82, 273 (1812).
1 Received June 15, 1928. Publication approved by the Director of 5 Rose, Ann. Physik (Pogg.), 94, 481 (1855); 96, 96, 284 (1855).
the Bureau of Standards of the U. S. Department of Commerce. 8 Cf. Nernst, Lehrbuch der theoretischen Chemie, 2 Aufl., p. 409
2 IND.ENG. CHEM.,20, 1234 (1928). 7 Malaguti, Ann. chim. phys., [3] 37, 198 (1853); 61, 328 (1857).
8 By A. T. McPherson, of the Bureau of Standards. 8 Meyerboffer, 2. physik. Chem., S3, 513 (1905).
30 ANALYTICAL EDITION Vol. 1, No. 1
not materially affect the principles involved, and that if the ing in order to prevent bumping. This might have been
procedure were available for the purpose mentioned it would avoided and the concentration of the solution kept nearly con-
also be applicable in other cases. stant by attaching a condenser to the flask, but it was con-
The actualprocedurein most caseswas to place 0.4to0.5gram sidered preferable to carry out the boiling as it would most
of pure sodium sulfate into a Pyrex Kjeldahl flask (of about likely be done in routine analytical work.
800 cc. capacity), add an equivalent or slightly larger amount The results are given in Table I.
of barium chloride solution containing 0.1 gram BaClz per Table I
cubic centimeter, and finally 3 cc. of 60 per cent perchloric EQUIVA- TIME
acid. A glass bead about the size of a small pea was also NaaSOi NarCOa
LENTSOP OB Bas04
TAKEN NarCOa CONCN.HEATINGOBTAINBD~NaaSOl FOUND
placed in the flask to prevent bumping. The flask was then Gram Hour Gram Gram Per cent
placed in an inclined position on a piece of asbestos gauze and 0.4067 4.9 2N 0.5 0.4445 0.2705 66.5
0.3996 7.8 2 N 0.5 0.6095 0.3709 92.8
heated to gentle boiling with a Bunsen burner until the 0,4085 12.0 2 N 0.5 0.6561 0.3992 97.7
water was nearly all driven off and perchloric acid fumes be- 0.4026 15.5 2 N 0.5 0.6564 0.3994 99.2
0.4061 6.0 4 N 1.0 0.6139 0.3736 92.0
gan to appear in the flask. 0.4098 9.0 4 N 1.0 0.6636 0,4037 98.5
0.4170 11.6 4 N 1.0 0.6718 0,4088 98.2
Since dilute solutions of perchloric acid lose very little of 0.4034 14.8 4 N 1.0 0.6602 0.4018 99.6
&heacid on boiling until their concentration has reached 60 0.4266 16.7 4 N 1.0 0.6984 0.4250 99.6
0.4400 19.4 4 N 1.0 0.7204 0.4384 99.6
per cent, and since, in the present work, the boiling was 0.4832
0.4878
23.5
29.1
4 N 1.0 0.7903 0.4809 99.5
4 N 1.0 0.8015 0.4877 100.0
stopped when the fumes of perchloric acid first appeared in Not corrected for occlusions.
the flask, it may be assumed that most of the acid originally
used was still present. I n no case, even in the last experiment, was the residue
To prevent this acid from interacting with the sodium of barium carbonate which was filtered off from the solution
carbonate subsequently added, and thus materially reducing completely soluble in hydrochloric acid. On the other hand,
the proportion of this reagent,, 4.6 cc. of a solution of sodium the results given have not been corrected for occlusions of
6 hydroxide (0.238 gram per cubic centimeter) were added, other salts by the barium sulfate, and thus two errors balance
this being equivalent to 99 per cent of the perchloric acid each other to a certain extent.
originally taken. The sodium carbonate solution was then When the sodium carbonate solution and barium sulfate
added in varying amounts and concentrations, and heated were heated together on a steam bath instead of over a
for different periods of time, either on a steam bath or to Bunsen burner, results similar to the above were obtained,
gentle boiling over a Bunsen burner. The solution was but approximately twice as much time was required for the
filtered twice (on account of the tendency of the barium same degree of conversion.
carbonate to climb over the top of the filter paper), the MODIFIEDPRocEDuRE-In the next series of experiments
filtrate acidified with hydrochloric acid, and the sulfate the procedure was slightly modified. Instead of neutraliz-
precipitated with barium chloride and determined in the ing the perchloric acid with sodium hydroxide, as had been
usual manner. done in previous experiments, the acid was first diluted with
Meyerhoffers and others recommend that the residue of 20 cc. of water, then 7.5 grams of powdered sodium carbonate
barium carbonate be washed first with a solution of sodium were added, and finally the neck of the flask was rinsed down
carbonate until sulfates are completely washed out and finally with an additional portion of 10 cc. of water. This amount
with water. This was not done in most cases in the present of carbonate was sufficient to neutralize the perchloric acid
work; instead, the solution from the Kjeldahl flask was first (3 cc. of 60 per cent solution) even if none of it had been ex-
allowed to run through the filter completely before washing pelled, and still leave 6 grams of the carbonate to react with
with water was started. This procedure was adopted for the barium sulfate. Furthermore, since the amounts of
several reasons. I n the first place, the bulk of the sodium sodium sulfate originally taken in this series of experiments
sulfate in the solution would have been removed from the were very close to 0.4 gram in each case, the amount of sodium
presence of the barium carbonate and the small amount re- carbonate left after neutralizing the perchloric acid was a t
maining could have only a slight effect upon the barium least 20 times that theoretically required to decompose the
carbonate in the short time during which i t would remain in barium sulfate. The heating was continued for 2 hours aft8er
contact with it. Secondly, although the small amount of the addition of sodium carbonate. I n the first three experi-
sodium sulfate would theoretically have some action on the ments, after the barium carbonate was filtered off, it was
residue, there would also be some sodium carbonate present washed first with 20 cc. of 4 N sodium carbonate, then with
in the same ratio as in the original solution; and although 20 cc. of a normal solution of the same salt, and finally with
the effect of this carbonate in the more dilute solution would water; whereas in the last two experiments the filter was
be much weakened, it would undoubtedly prevent the sodium washed with water only. The results are given in Table TI.
sulfate from exerting its full influence, particularly since the Table I1
sulfate would also be diluted in the same proportion as the NazSO4 Bas04
carbonate. Finally, it was desired to know the efficiency of EXPT. TAKBN OBTAINED~ NaaSO4 FOUND=
Gram Gram Gram Per cent
this method of extracting sulfur from barium sulfate under 1 0.4008 0.6556 0.3990 99.5
the least favorable conditions. 2 0.4005 0.6554 0.3988 99.6
3 0.4048 0.6624 0.4032 99.6
4 0,3986 0.6510 0.3962 99.4
Experimental Procedure 5 0.3984 0.6525 0.3970 99.6
a Not corrected for occlusions.
PRELIMINARY
EXPERIMENTS-In a series of preliminary
experiments varying quantities of a 2 N or 4 N solution of The barium carbonate remaining on the filter was then
sodium carbonate (0.106 or 0.212 gram per cc.) were added to treated with dilute hydrochloric acid and the insoluble resi-
the barium (sulfate) in the Kjeldahl flasks and the mixture due filtered off, washed, ignited, and weighed. The sulfate
was heated to gentle boiling for 0.5 or 1 hour, as indicated. in the residue was then determined by fusion with sodium
(Table I) The carbonate solution was diluted somewhat in carbonate (1gram) in the usual manner.
each case by the solution already in the flask, but during the The weight of the insoluble residue from the barium car-
boiling it gradually became concentrated so that it was neces- bonate and the weight of the barium sulfate obtained from
sary sometimes to add some water before the end of the heat- the residue after fusion with sodium carbonate are shown in
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 31
Table 111. The first column gives the number of t,he experi- carbonate and boiled for 2 hours, after which the resulting
ment corresponding with the number given in Table 11, and solutions were filtered, acidified, and precipitated with barium
the last column gives the proportion of barium sulfate chloride. The results are given in Table IV.
(recovered from the residue after fusion) as percentage of When the barite was boiled with the sodium carbonate
the total, original barium sulfate obtained from the sodium solution for only 1hour, the proportion of precipitated barium
sulfate. sulfate averaged only about 98.5 per cent of the original ba-
Table 111
rite. The same barite when decomposed with fused sodium
RESIDUE Bas04 PROPORTION carbonate gave an average of 99.93 per cent of barium sulfate
FROM FROM OF ORIQINAL precipitate.
EXPT. BaCOi RESIDUE BaSO'
Gram Gram Per cent
1 0.0051 0.0047 0.71 Conclusions
2 0.0047 0.0040 0.61
3 0.0048 0.0042 0.03 It is obvious that the removal of sulfur from barium sulfate
4 0.0050 0.0045 0.09
5 0.0053 0.0046 0.70 by digestion with hot solutions of sodium carbonate does not
The results given in the last two tables, as compared with yield such good results for analytical purposes as does the
the preceding, show that nothing is to be gained by in- fusion method. Nevertheless, because of its simplicity and
creasing the proportion of the sodium carbonate beyond speed, especially when combined with a method like the
fifteen times the theoretical, or by heating beyond 1 hour. nitric-perchloric acid method of determining sulfur in rubber,
While i t is probable that equilibrium has not yet been reached, this procedure may find some usefulness where extreme ac-
the reaction a t this point is so slow that it is not practicable curacy is not so important as speed; and if only a small pro-
to attempt to force i t further. It would also appear that the portion of barium sulfate is present-15 per cent or less-it
importance of first washing the barium carbonate (after may yield results well within the limits of experimental error.
filtering) with sodium carbonate, in order to remove all soluble To adapt this method of decomposing barium sulfate t o
sulfates, has been overestimated, and that no appreciable the determination of total sulfur in rubber by the nitric-per-
error will be introduced by washing immediately with water. chlorio acid method,2 the procedure would be as follows:
This will be especially true if the original solution is allowed After dissolving the rubber sample in the mixture of nitric
to run completely through the filter before wa,shingis started. and perchloric acids and boiling until fumes of perchloric
acid appear, allow the solution to cool somewhat, add about
Table IV
BARITETAKEN B a s 0 1 PRECIPITATED"
25 cc. of water, then 5 grams of sodium Carbonate, and boi1
Gram Gram Per cent gently for 1 hour (or for 2 hours if native barite is present or
0.5005 0.4980
0.4971
99.5
99.7
suspected), adding more water from time to time if necessary
0.4986
0,6005 0.4985 99.6 to prevent bumping. The solution is then filtered, acidified,
0.5002 0.4977 99.5 and treated with barium chloride in the same manner as in
a Not corrected for occlusions.
the fusion method.
EXPERIMENT8 WITH BARITE-Barite, the natural barium The amount of sodium carbonate recommended above
sulfate, will give very much the same kind of results as the (5 grams) is sufficient to neutralize 3 cc. of 60 per cent per-
above, although it requires longer heating with the sodium chloric acid and still leave fifteen times the amount of sodium
carbonate solution because it is usually less finely divided and carbonate theoretically required to decompose 0.5 gram of
hence presents a relatively smaller surface than the arti- barium sulfate-that is, enough for a 1-gram sample of a
ficially precipitated material. Half-gram samples of native rubber compound containing 50 per cent of barium sulfate.
barite, ground to 200 mesh, were boiled with about 25 cc. of While this reaction was studied especially with a view of its
water and 3 cc of perchloric acid as in previous experiments, utilization in the analysis of rubber, it may, of course, find
then again treated with 25 cc. of water and 6 grams of sodium applications elsewhere.
T HAS long been apparent to the author that a definition decomposition seems to be assured by the fact that samples
T HAS long been apparent to the author that a definition decomposition seems to be assured by the fact that samples
Evaluation of Stibnite
I-De termination of Sulfur
Wallace M. McNabb and E. C. Wagner
UNIVERSITY
ox PENNSYLVANIA,
PHILADELPHIA,
PA.
ETHODS for the analysis of stibnite intended for tendency for the evolution method to yield results for sulfur
M use in primers were described in 1918 by Cushman,2 slightly lower than those obtained by bromine oxidation.
and were incorporated into the specifications of the Table I-Comparison of Sulfur Determinations by Evolution a n d
United States Ordnance D e ~ a r t m e n t . ~I n the recommended
EVOLUTION
Bromine Oxidation Methods
METHOD BROMINE OXIDATION
method for sulfur this element is oxidized by action of bro- % %
mine and glacial acetic acid, and after suitable further treat- 21.14 21.38
21.15 21.28
ment, including removal of antimony by use of aluminum 21.14 21.40
powder, it is precipitated and weighed as barium sulfate. 21.10
Av. 2 1 . 1 3 Av. 2 1 . 3 4
This procedure, Cushman states, was adopted a t the Frank- 21.72 21.77
ford Arsenal as giving fairly quick and accurate results. 21.72
21.70
21.78
Elsewhere in the same article occurs the declaration: All 21.76
other methods for the determination of sulfur have been re- 21.73
Av. 2 1 . 7 3 Av. 2 1 . 7 8
jected as leading to inconsistent results. This condemna-
tion specifically includes fusion methods, evolution methods, The differences of 0.21 and 0.05 per cent appeared to be
and electrolytic methods, and is offered without supporting significant, and not accidental. As the bromine oxidation is
\data. clearly a method for total sulfur, and as the evolution method
Wofk on the analysis of stibnite in 1917 showed the ap- can determine only sulfide (and oxysulfide) sulfur, it was
Tplicability of the evolution method, the sulfur being disen- suspected that the variations might be due to the presence
gaged as hydrogen sulfide, absorbed in ammoniacal cadmium of free sulfur in the stibnite. To test the two methods with
solution, and determined iodometrically by an adaptation respect to their ability to indicate a distinction between total
of the familiar procedure. There was noticed a constant sulfur and sulfide sulfur, a specimen of artificial black anti-
mony sulfide was analyzed for sulfur by both methods, and a
1 Received July 11, 1928.
Cushman, J. IND.ENG.CHEM.,10,376 (1918).
10-gram sample was extracted with carbon tetrachloride and
a U. S. Army Ordnance Dept., Sfiec. 60-11-14(July 24,1923). the free sulfur weighed.
32 ANALYTICAL EDITION Vol. 1, No. 1
The use of nearly 60 grams of combined dish, sand, and rod and after 4 hours heating the residue showed no indication
and the many weighings necessary to determine moisture in whatever of blackening or decomposition. While a small
1 gram of sirup make a cumbersome task and it was with amount of water continued to distil over, the major part
great pleasure that the author welcomed the method of Bid- was over in 1 hour and the slight additional amount a t the
well and Sterling.* end of 3 hours made the total in close agreement with the
As soon as apparatus could be made the method was tried percentage as indicated by the official method. As an indi-
out, but with only partial mccess because the same continu- cation of the effect of the Filter-Cel it was noted that the
ous decomposition was encountered and no definite end same rate of distillation of the toluene could be maintained
point was found. Several tests showed that by using iden- with the oil bath at 127 C. as a t 143 C. without Filter-Cel.
tical conditions and a definite time interval concordant re- The advantages of the distillation method as set forth by
sults were possible on sirups of moderate or low invert sugar Bidwell and Sterling were found to be true, especially the
content, but with invert sugar of 25 per cent or over the re- advantage of a 15-gram sample and a single weighing. The
sults were unreliable and the residue in the flask after dis- total time consumed is a t most 4 hours, during which the
tillation was seriously blackened. At about this stage an operator is occupied but a few minutes at the start.
oil bath was substituted for a sand bath with improved re- The most satisfactory manipulation is to put half of a
sults, but the method was considered as limited to black- roughly weighed 10-gram portion of dried Filter-Cel into a
straps and sirups of low invert sugar content. flask and then weigh off 15 grams of sirup into a capsule
With the thought in mind to precipitate any calcium formed from a piece of thin waxed paper resting in a 15-cc.
present as chloride and that some molecular rearrangement Gooch crucible. After weighing, the paper may be lifted
might cause a more ready release of water present, a dis- out of the crucible and sirup and all dropped into the flask,
tillation determination was tried with 2 grams of powdered after which the second half of the Filter-Cel is added and
sodium oxalate added to 15 grams of sirup. The result was the distillation carried out as described by Bidwell and
a quicker distillation and less blackening of the residue. Sterling.
Another test with 10 grams of sodium oxalate gave results
Comparison of Distillation M e t h o d w i t h Official Method for
that were even better. Inasmuch as blackstrap has a con- D e t e r m i n a t i o n of Moisture in Sugar Sirups
siderable quantity of suspended particles and sodium oxalate 3 HOURS VACUUM
SAMPLE BATR FILTER-C&tDISTILLATIONOVEN INVERT
is only sparingly soluble, the question arose as to whether the Grams Per cent Per cent Per cent
improved results were due to the chemical or physical effect A Sand 1
of the sodium oxalate. To determine if finely divided ma- B
Oil
Sand 1
terial would assist in the distillation, 10 grams of Filter-Cel Oil
were added. The result was rapid evolution of the water, Oil IU
Evaluation of Stibnite
I-De termination of Sulfur
Wallace M. McNabb and E. C. Wagner
UNIVERSITY
ox PENNSYLVANIA,
PHILADELPHIA,
PA.
ETHODS for the analysis of stibnite intended for tendency for the evolution method to yield results for sulfur
The differences between results by the two methods average Determination of Sulfur by Bromine-Acetic Acid Method
0.65 per cent, which is very nearly the amount of free sulfur
found. The directions given by Cushman2 were used unaltered.
That the bromine-acetic acid method yields results for the This analysis, if started late in the afternoon and the oxida-
total sulfur of stibnite was shown further by trials using this tion allowed to proceed overnight, can scarcely be com-
procedure in comparison with the carbonate fusion method pleted before late the following afternoon. The results (for
of F r e ~ e n i u s . ~The results for the first specimen of stibnite total sulfur) are concordant, and appear to be good, though
mentioned above were 21.38 and 21.31 per cent of sulfur some question might be raised as to possible interference of
whose 21.35 per cent mean is in excellent agreement with the the added aluminum with the precipitation of barium sulfate.6
21.34 per cent found by bromine oxidation. The method of Determination of Sulfur by Evolution Method
Fresenius, though found to be applicable (it is among those
rejected by Cushman), is fully as inconvenient as the official The sample of stibnite is decomposed by hot concentrated
method, and its results have the same significance. hydrochloric acid in an evolution flask, whereby sulfur present
Table 11-Comparison of Results by Two Methods on Sample Known as sulfide and oxysulfide is given off as hydrogen sulfide, while
to Contain Free Sulfur any free sulfur is unaffected. The hydrogen sulfide is boiled
BROMINE OXIDATION EVOLUTION METHODCCla EXTRACTION out, and with the aid of a stream of carbon dioxide (or hydro-
SULFUR SULFUR F R ESULFUR
~
% % % gen) is conveyed into a receiver containing ammoniacal cad-
26.16 25.48 0.70 mium solution. The precipitate of cadmium sulfide is filtered
26.17 25.53 0.74
25.53 off, washed, and decomposed by hydrochloric acid in contact
25.62 with a measured excess of standard iodine solution, of which
25.45
Av. 26.17 Av. 25.52 Av. 0.72 the excess is finally titrated with standard thiosulfate solution.
The acid liquid in the evolution flask contains the antimony
Recently there have been investigated the actual presence as trichloride, and may be used for the iodometric determina-
of free sulfur in the Chinese stibnite used in primers, its in- tion of the metal.
fluence upon the determination of sulfur by bromine oxida- APPARATUS-The evolution apparatus recommended for
tion, and also the amount of the error introduced by the small analysis of strong oxidizing agents by Bunsens method8 may
quantity of sulfate sulfur usually present. The evolution be used, as could also, no doubt, that designed and described
method was subjected to critical trial and improvement, as by Scott.B The simpler arrangement shown in the sketch
reported below. The determination of antimony will be is fully as satisfactory, and may be assembled from ordinary
considered in a later paper. laboratory apparatus.
The occurrence of free riultur in stibnite, though not men-
tioned by Cushman, apponrs upon inquiry to be fairly com-
mon. Specimens of stibnite have been described6 with free
sulfur present, sometimes in well-formed crystals. The
specimens of primer stibnite used in the present study all
contained sulfur extractable by carbon tetrachloride.
Below are described the methods employed, and the results
obtained, in the examination of four lots of stibnite for sulfur
by bromine oxidation and by evolution, for free sulfur by
extraction with carbon tetrachloride, and for sulfate sulfur.
Determination of Free Sulfur
Transfer 10 grams of powdered stibnite to an extraction
thimble, and extract with purified carbon tetrachloride in a
Soxhlet apparatus. After 6 to 8 hours evaporate the sol-
vent and weigh any residue of sulfur.
I n trials of this method the extractions were continued dur-
Separatory funnel A contains concentrated hydrochloric
ing a further interval, but no increase in the weight of the
extraction flask occurred. The carbon tetrachloride was acid. Flask B is charged with lumps of calcite covered with
shown by blank tests to be quite free from non-volatile water. Drechsel bottle C contains a strong solution of pyro-
matter. Four specimens of stibnite thus examined yielded gallol and sodium bicarbonate, and D contains water. The
0.077, 0.09, 0.107, and 0.065 per cent of free sulfur. The small separatory funnel E is a t first empty. The decomposi-
extracted material was in one case identified by oxidation with tion flask F is a 250-ml. Erlenmeyer flask; G is a bulbed tube
bromine in carbon tetrachloride and precipitation as barium which serves as a partial condenser, to reduce the danger that
sulfate, 0.060 per cent of the 0.065 per cent originally weighed antimony chloride may distil over. Receiver H i s a 250-ml.
being recovered in this form. Erlenmeyer flask, charged with ammoniacal cadmium solu-
tion. It is surrounded by cold water contained in beaker I .
Determination of Sulfate Sulfur A second receiver was present in most of the experimental
trials, but was shown to be unnecessary. The stopper of
Boil 10 grams of the powdered stibnite for several niinutes the flask P is of rubber.
with 100 ml. of distilled water, and filter. Acidify the filtrate The stream of carbon dioxide may be replaced by one of
with hydrochloric acid, add barium chloride solution, and hydrogen, generated from zinc and hydrochloric or sulfuric
after about 6 hours filter the precipitate on paper and deter- acid. In this case washing bottle C must be charged with
mine as usual. alkaline lead solution.
The four stibnites examined yielded the following results
for sulfate sulfur: 0.011, 0.012, 0.010, and 0.012 per cent. Treadwell-Hall, 09. cit., p. 403.
7 Cf. Scott, Standard Methods of Chemical Analysis, Vol 1, p. 500
4 Treadwell-Hall, Analytical Chemistry, 6th ed., Vol. 11, p. 316. (1922).
Foshag, U. S. Geol. Survey, Bull. 796-8, p. 116; Eakle, 2. Kryst. 8 Wagner, IND. ENQ.CHEW.,16, 616 (1924).
Mineral., 24,587 (1895). Q Scott, 09. cit., p. 501.
34 ANALYTICAL EDITION Vel. 1, No. 1
There is required also a suction filter provided with a sup- Place the filter and precipitate in a 600-ml. beaker, and
porting cone. cover with 200 ml. of water. Add from a pipet 50 ml. of
SOLUTIONS-Sodium Thiosulfate. A solution of about 0.12 standard iodine solution, and introduce gradually and with
N strength is most convenient. This solution is accurately stirring 40 ml. of concentrated hydrochloric acid. Break
and easily standardized as follows, using potassium iodate as up lumps of precipitate and continue gentle stirring until no
primary standard?' undecomposed flocks of sulfide are visible when the liquid is
Weigh out accurately 0.8918 gram of pure dry potassium allowed to come to rest and the beaker viewed from beneath.
iodate. The best c. P. grade is satisfactory, but is slightly The liquid is now turbid due to separated sulfur, and must be
improved by recrystallization from water. Dissolve the decidedly brown or yellow due to excess iodine. Titrate the
iodate in water, transfer to an accurate 250-ml. volumetric iodine with standard thiosulfate solution, adding 5 ml. of
flask, and dilute to the mark a t 20' C. As required, with- 0.5 per cent starch indicator when the yellow color becomes
draw measured portions of this exactly decinormal solution faint. The end point is the disappearance of the last tinge
by means of a 25-ml. pipet calibrated against the flask. of blue-gray color.
Transfer to a 250-ml. Erlenmeyer flask 2.5 grams of iodate- STANDARDIZATION O F IODINE SOLUTION-Transfer to a 600-
free potassium iodide, dissolve in 25 ml. of water, and add ml. beaker 200 ml. of water and half of a filter paper. With
5 ml. of 6 N hydrochloric acid. As promptly as possible intro- the pipet used before introduce 50 ml. of standard iodine
duce from a pipet 25 ml. of the decinormal iodate solution solution, and then add gradually 40 ml. of concentrated hydro-
and a t once dilute the solution to 150 ml. Titrate the liber- chloric acid. Allow to stand, with stirring; for the same time
ated iodine with the thiosulfate solution, using as indicator required to decompose the cadmium sulfide in the analysis,
5 ml. of 0.5 per cent solution of soluble starch. Conduct a and titrate with standard thiosulfate.
blank titration in the same way, not omitting a like exposure
of the acidified potassium iodide solution to the air before COMMENTS-(~)A complete blank analysis is an excellent
precaution against error caused by impure reagents. One series
dcution. The blank titration is usually zero (no color with of trials yielded persistently high results, which by an over-all
stmch), but it should always be made. blank were finally shown to be due to hydrogen sulfide present
lodine. A solution of about 0.11 N strength is most con- in the hydrochloric acid.
venient--14 grams of iodine and 40 grams of iodate-free (2) The treatment of the cadmium sulfide precipitate on the
potassium iodide per liter." This solution is standardized filter, as prescribed above, appears to be necessary. Failure to
bring the precipitate to a dense and caked condition by final
against the thiosulfate under' the conditions of the analysis, application of strong suction was associated with variable but
as described later, high results, the error amounting occasionally to 0.8 per cent of
Ammoniacal Cadmium Solution. Dissolve in 500 ml. of sulfur.
water 63 grams of cadmium chloride or 74 grams of the sul- (3) Too vigorous or too prolonged boiling will result in some
fate, add 1250 ml. of strong ammonia water (sp. gr. 0.90), distillation of antimony chloride. This of course affects results
for antimony, but has not been found definitely to influence those
and dilute to 2500 ml. Of this solution 50 ml. suffice for for sulfur.
0.184 gram of sulfur. (4) Permanent absorption of iodine by a t e r paper under the
Procedure-Use for the analysis material which has been conditions of analysis appears not to occur. Thus 50 ml. of iodine
ground and passed through a 100-mesh sieve. required, in absence of iilter paper, 43.02 and 43.05 ml. of thio-
Weigh out 0.2 to0.25 gram of the stibnite,and transfer tothe sulfate. In presence of filter paper, and under the conditions
described, 50 ml. of iodine required 43.05 and 43.01 ml. of thio-
flask F of the apparatus. Charge the receiver H with 100 sulfate.
ml. of ammoniacal cadmium solution and 50 ml. of water. (5). The use of EL rubber stopper in the decomposition flask is
Pass through the apparatus a stream of carbon dioxide (or permissible. Comparative trials using the rubber-stoppered flask
hydrogen) until removal of air is assured; this requires about and an all-glass evolution apparatus yielded identical results.
5 minutes. Introduce 30 ml. of concentrated hydrochloric Analytical Results
acid into the addition funnel E, and allow it to enter the flask.
With the current of gas passing steadily, heat to very gentle The results for four specimens of primer stibnite, examined
boiling the acid in the decomposition flask, and continue the by the methods discussed above, are given in Table 111.
gentle boiling until the sample is decomposed; this requires Table 111-Analytical Data
10 to 15 minutes. When decomposition is complete, re- EVOLUTIONCClr EXT. BROMINE
move the flame, and continue the passage of carbon dioxide METHOD FREE SULFATE TOTAL OXIDATION
SULFUR SULFUR SULPUR SULFUR TOTAL
for about 20 minutes longer. .--,
[A) . .
(B) (Cl
. . (A + +B C) SULFUR
The receiver now contains all the sulfide sulfur of the sample % % % % %
20.05 20.16
in the form of cadmium sulfide. The liquid in the decomposi- 20.03 20.14
tion flask contains the antimony as trichloride, and may be 20.08 20.18
20.02
used for the determination of this metal. 20.08
Av. 20.05 0.08 0.01 20.14 20.16
Disconnect receiver H , remove the stopper and inlet tube,
and wash the latter free from any adhering precipitate.
Allow the cadmium sulfide to settle, and decant the liquid
through a suction filter supported by a cone. The filter
paper should be of good quality (not affected by dilute iodine 21.66
solution); Munktell No. 00, of 90 mm. diameter, was used 21.60 21.71
21.64 21.66
and is satisfactory. Wash the cadmium sulfide several times 21.65 21.73
21.66
with water by decantation, and then transfer the precipitate Av. 21.64 0.11 0.01 21.76 21.70
to the filter. Wash further several times with water, applied
in a strong stream so as to churn up the precipitate thoroughly
each time. Finally apply suction strongly, SO that the
precipitate becomes densely caked on the paper. Open the
filter, and tear off and discard the portions which are entirely
free from cadmium sulfide. The results obtained by the evolution method fail to sup-
10 Cf. Treadwell-Hall, 09. cit., p. 553.
port Cushman's condemnation of this procedure. It is
11 Chapin, J . Am. Chem. Soc., 41, 357 (1919). indicated to be a method for sulfide sulfur, whereas bromine
January 15, 1929 I N D U S T R I A L A N D ENGINEERING CHEMIXTRY 35
oxidation yields results for total sulfur, including sulfide determination of both sulfur and antimony in a single sample,
sulfur, sulfate sulfur (negligible), and free sulfur. The demon- both analyses being iodometric.
stration that primer stibnite contains free sulfur extractable
with carbon tetrachloride seems to introduce a new considera- Summary
tion into the analysis of this mineral. If the amounts of free Commercial stibnite of the quality required for primer
sulfur in the specimens examined may be considered repre- manufacture is likely to contain free sulfur. This can be de-
sentative, then the error admitted when this sulfur is termined by extraction with carbon tetrachloride.
calculated to SbzSs will in general be of the order of 0.4 The bromine-acetic acid method for determination of sul-
per cent, and it could of course be much larger without fur yields results for total sulfur, and not specifically for
detection by the official method. The presence of free sulfide sulfur. Its results are therefore not a correct basis
sulfur in stibnite to be used in primers is probably not for calculation of the apparent Sbz& content of stibnite.
harmful, but the impropriety of calculating it to SbzSa is The evolution procedure described above is recommended
obvious. as a rapid and satisfactory method for determination of the
The sulfide sulfur of stibnite includes that present in sul- sulfide sulfur of stibnite.
fides of antimony, lead, iron, and perhaps arsenic, the last Acknowledgment
three being commonly present as impurities. According to
Cushman,2 iron and lead sulfides are not objectionable im- The writers are indebted to H. C. Pritham, director of
purities, and may even be used as fairly good substitutes for laboratories of the Frankford Arsenal, for the specimens of
stibnite in primers. primer stibnite used in this work, and to F. L. Hess, of the
As regards rapidity and convenience, the evolution pro- Bureau of Mines, for information regarding the occurrence of
cedure is much superior to the official method, and permits free sulfur in native stibnite.
FTEN it is impossible to observe the falling ball in tlie constants of which are shown in the diagram. The coils
0 viscosity measurements owing to the color of the solu-
tion being tested. This is especially true of cupram-
monium solutions of cellulose. If the cellulose is dirty, the
of t<hecircuit are wound in two slots on the viscosity tube as
in Figure 2. The slots of pasteboard are spaced so that their
centers are separated by the distance over which the fall of
solution is usually opaque and even the strongest light cannot the ball is to be measured. The grid coil is wound first, half
penetrate it. The following method has been successfully being wound in tlie first slot and half in the second. The
applied on this type of solution. It has the advantage of plate coil is then wound half in each slot directly over the
substituting steel balls easily duplicated in dimensions for grid coil. There are usually about seventy-five turns of No,
the glass beads which are rather difficult to duplicate. 28 wire in each slot for the grid coil and about fifty in each
Apparatus slot for the plate coil.
The radio vacuum tube has been applied to this method. Method
I The circuit (Figure 1) consists of a simple oscillating circuit, With the coils of the viscosity tube correctly hooked to the
1 Received September 20, 1928. apparatus, it is set
in oscillation by
turning on the fila-
ment of the vac-
uum t u b e . The
resistance of the
grid leak is slowly
decreased from the
maximum until the
frequency of oscil-
lation gives the de-
sired tone. The
frequency can also
be controlled to a
limited extent by
v a r y i n g the fila-
ment rheostat.
W i t h t h e s e t so
oscillating as to
give t h e desired
phones tone a steel ball is Figure 2-Viscosity Tube and Attached
Figure 1-Circuit for Vacuum Tube dropped t h r o u g h Vacuum Tube Circuit
January 15, 1929 I N D U S T R I A L A N D ENGINEERING CHEMIXTRY 35
oxidation yields results for total sulfur, including sulfide determination of both sulfur and antimony in a single sample,
sulfur, sulfate sulfur (negligible), and free sulfur. The demon- both analyses being iodometric.
stration that primer stibnite contains free sulfur extractable
with carbon tetrachloride seems to introduce a new considera- Summary
tion into the analysis of this mineral. If the amounts of free Commercial stibnite of the quality required for primer
sulfur in the specimens examined may be considered repre- manufacture is likely to contain free sulfur. This can be de-
sentative, then the error admitted when this sulfur is termined by extraction with carbon tetrachloride.
calculated to SbzSs will in general be of the order of 0.4 The bromine-acetic acid method for determination of sul-
per cent, and it could of course be much larger without fur yields results for total sulfur, and not specifically for
detection by the official method. The presence of free sulfide sulfur. Its results are therefore not a correct basis
sulfur in stibnite to be used in primers is probably not for calculation of the apparent Sbz& content of stibnite.
harmful, but the impropriety of calculating it to SbzSa is The evolution procedure described above is recommended
obvious. as a rapid and satisfactory method for determination of the
The sulfide sulfur of stibnite includes that present in sul- sulfide sulfur of stibnite.
fides of antimony, lead, iron, and perhaps arsenic, the last Acknowledgment
three being commonly present as impurities. According to
Cushman,2 iron and lead sulfides are not objectionable im- The writers are indebted to H. C. Pritham, director of
purities, and may even be used as fairly good substitutes for laboratories of the Frankford Arsenal, for the specimens of
stibnite in primers. primer stibnite used in this work, and to F. L. Hess, of the
As regards rapidity and convenience, the evolution pro- Bureau of Mines, for information regarding the occurrence of
cedure is much superior to the official method, and permits free sulfur in native stibnite.
FTEN it is impossible to observe the falling ball in tlie constants of which are shown in the diagram. The coils
0 viscosity measurements owing to the color of the solu-
tion being tested. This is especially true of cupram-
monium solutions of cellulose. If the cellulose is dirty, the
of t<hecircuit are wound in two slots on the viscosity tube as
in Figure 2. The slots of pasteboard are spaced so that their
centers are separated by the distance over which the fall of
solution is usually opaque and even the strongest light cannot the ball is to be measured. The grid coil is wound first, half
penetrate it. The following method has been successfully being wound in tlie first slot and half in the second. The
applied on this type of solution. It has the advantage of plate coil is then wound half in each slot directly over the
substituting steel balls easily duplicated in dimensions for grid coil. There are usually about seventy-five turns of No,
the glass beads which are rather difficult to duplicate. 28 wire in each slot for the grid coil and about fifty in each
Apparatus slot for the plate coil.
The radio vacuum tube has been applied to this method. Method
I The circuit (Figure 1) consists of a simple oscillating circuit, With the coils of the viscosity tube correctly hooked to the
1 Received September 20, 1928. apparatus, it is set
in oscillation by
turning on the fila-
ment of the vac-
uum t u b e . The
resistance of the
grid leak is slowly
decreased from the
maximum until the
frequency of oscil-
lation gives the de-
sired tone. The
frequency can also
be controlled to a
limited extent by
v a r y i n g the fila-
ment rheostat.
W i t h t h e s e t so
oscillating as to
give t h e desired
phones tone a steel ball is Figure 2-Viscosity Tube and Attached
Figure 1-Circuit for Vacuum Tube dropped t h r o u g h Vacuum Tube Circuit
36 ANALYTICAL EDITION Vol. 1, No. 1
the solution, and when the ball passes through the coils a Both cuprammonium solutions of cellulose and nitrocellulose
change in the note is observed due t o the change in frequency solutions in various solvents have been run by the method
of oscillations. Since the coils are identical, the time elaps- with satisfactory results.
ing between the change in hum a t the upper and lower coils The following table gives an idea of the accuracy of the
is the time required by the ball to pass through the distance method when used to measure viscosities of cellulose in
being measured. cuprammonium solutions and using steel balls of various sizes:
Conclusions
I/S-INCHBALL 8/16-INCA BALL I/I-INCEI BALL
Seconds Seconds Seconds
The apparatus has several limitations. If the solution is 14.0 8.0 5.8
too viscous, the change in note may be so gradual as to be 14.6 8.0 6.8
14.6 8.0 5.8
inaudible to any but the trained observer. The apparatus 14.8 8.0
works best with steel balls above 1/8 inch (3 mm.) in diameter. 14.6 Duplicate soln.
14.0 8.2
Smaller balls do not exert enough influence on the coils t o 8.4
8.0
change the note sufficiently for accurate audible detection. 8.2
T HOSE who have made any large number of accurate have proved so satisfactory as t o accuracy and economy of
vapor pressure determinations fully appreciate the time, labor, and materials that they are thought worthy of
faults of most all methods, both static and dynamic, being passed on to other possible users.
which have been published. The static methods are un- Apparatus
desirable either because of inherent sources of inaccuracy or
because of the excessive time necessary to get results. Some Figure 1 illustrates the apparatus described by Ramsay
of the dynamic methods offer possibilities of high accuracy, and Young, while Figure 2 shows the present writers modi-
but they are time-consuming, and those that offer speed do fication of the part ACE, Figure 1. The main reason for
so a t the sacrifice of ac- modifying Ramsay and Youngs apparatus was to prevent
a c u r a c y . A good r6- distillation of the sample under test from A into C, where it
sum6 of vapor pressure would condense with a consequent frequent interruption of
methods published up tests. With the apparatus modified as in Figure 2, all dis-
to 1926 is g i v e n b y tillation from A was condensed and returned a t a slow,
R e i 11y , R a e , a n d uniform rate, such as would never exert any disturbing effect
WI Wheeler.2 on the vaDor-liauid eauilibrium relationshitx existinn in
0 ~~-A
~~
T HOSE who have made any large number of accurate have proved so satisfactory as t o accuracy and economy of
vapor pressure determinations fully appreciate the time, labor, and materials that they are thought worthy of
faults of most all methods, both static and dynamic, being passed on to other possible users.
which have been published. The static methods are un- Apparatus
desirable either because of inherent sources of inaccuracy or
because of the excessive time necessary to get results. Some Figure 1 illustrates the apparatus described by Ramsay
of the dynamic methods offer possibilities of high accuracy, and Young, while Figure 2 shows the present writers modi-
but they are time-consuming, and those that offer speed do fication of the part ACE, Figure 1. The main reason for
so a t the sacrifice of ac- modifying Ramsay and Youngs apparatus was to prevent
a c u r a c y . A good r6- distillation of the sample under test from A into C, where it
sum6 of vapor pressure would condense with a consequent frequent interruption of
methods published up tests. With the apparatus modified as in Figure 2, all dis-
to 1926 is g i v e n b y tillation from A was condensed and returned a t a slow,
R e i 11y , R a e , a n d uniform rate, such as would never exert any disturbing effect
WI Wheeler.2 on the vaDor-liauid eauilibrium relationshitx existinn in
0 ~~-A
~~
I I I ii
liquid are then poured mentioned error due to air dissolved in the mercury, so that
into the funnel, which it is useless to eliminate the latter completely. For compari-
is stopped loosely to son of gasolines the ratio of vapor to liquid volumes must be
avoid excessive evapo- set arbitrarily, and this adjustment is particularly convenient
ration. The funnel is with this apparatus. The ratio might be 1:9, corresponding
lowered again with the to a tank 90 per cent full. For fairly pure liquids, such as
I I stopcockopen, and then commercial acetone, ether, etc., a much higher ratio, five, to
shut off with volatile ten times the liquid volume, would be preferable. Results
liquid on both sides of agreeing well with literature values have been obtained for
the stopcock. Again ether, acetone, benzene, carbon tetrachloride, chloroform,
the funnel is raised and and pentane. The time for a test is 5 to 10 minutes; for
the new height of mercury (a) read as before. The difference check determinations, less than 1 minute.
HE specific gravity of semi-solid bituminous substances both ends and cut longitudinally in two equal parts, which
T is usually determined by means of a wide-mouth are held together with a rubber band. The mold can hold
pycnometer or by the Sommer hydrometer. I n either about 40 grams of asphalt or pitch. It is placed on an
case the solidified material has t o be removed after each de- amalgamated brass plate and the molten sample poured into
termination, so that the cup holding the sample may be it.
cleaned and made ready for the nest test. GLASSCYLINDER-This cylinder is 8'/2 inches (21.6 cm.)
The cleaning of the apparatus usually consumes more time high and 13/4 inches (4.4 cm.) in diameter, open a t the top.
than the actual determination and causes considerable slop- At the bottom a three-hole rubber stopper is inserted, from
piness. Moreover, the delicate glass pycnometers very often which one connection is made to a 500-cc. separatory funnel,
, break during cIeaning, and with the Sommer hydrometer serving as a leveling bottle. Another connection leads to a
difficulties are frequently encountered in the removal of 50-cc. buret, the lower end of which has been cut off below
the cover and flange from the metal cylinder, as a result of the zero mark. Both the buret and the leveling bottle are
which the cylinder gets scratched or indented. held a t the desired height by
The writer has developed a sho;t method eliminating the means of an iron support. Into
cleaning of the pycnometer and requiring only one weighing. t h e t h i r d hole of the rubber
This method has been used for over three years in this labora- stopper is inserted a glass tube, E
tory for determining the specific gravities of asphalts employed '/4 inch (6 mm.) inside diameter,
for waterproofing purposes in the subways and ranging in leading to the outside of the
melting point from 125" to 175" F. (51.7" to 79.4" C.) (ball c y l i n d e r , parallel to its side. M
and ring), and has been found to give very satisfactory re- This glass tube has a mark etched
sults. Duplicate analyses made on the same material have on it, 5 inches (12.7 cm.) from
been found to check within 0.002. The method can be used the lower end. The glass cylin-
to advantage in connection with all kinds of solid and semi- der with the side-tube attach-
solid materials. ment may be clamped in another
A detailed description of the method and the apparatus iron support.
required is given below. I n the method described the specific This apparatus is not expen-
gravity is obtained at 77" F. (25" C.), which is the tempera- sive and can be constructed in
ture specified in the A. s.T. M. Standard Method of Test for any laboratory.
Specific Gravity of Asphalts and Tar Pitches, Serial Designa-
tion D71-27. Preparation of Sample
Description of Apparatus The sample of the bituminous
substance is melted in a tin can
BRASSMom-This consists of a brass tube, 13/4 inches at the lowest possible teppera-
(4.4 cm.) high and ls/16 inches (3.3 cm.) in diameter, open at ture and poured into the brass , , ~ ~ ~ ~ ~ ~ ~ "
1 Received October 30, 1928. mold to fill it slightly more than Bituminous Substances
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 39
A
4 cm. below the barom- in height gives the vapor pressure of the liquid a t the sur-
eter reading, b e c a u s e , rounding temperature. The correction for height of the
of dissolved air, vapors, liquid column (a-b) is within experimental error. For other
etc. T h e constant temperatures a jacket of warm water or steam, etc., around
height is taken as the the whole stem of the funnel must be provided.
zero reading. With impure liquids, such as gasoline, the observed vapor
A few cubic centi- pressure as determined by any method varies greatly accord-
meters of the volatile ing to the volume of the vapor space, far more than the above-
I I I ii
liquid are then poured mentioned error due to air dissolved in the mercury, so that
into the funnel, which it is useless to eliminate the latter completely. For compari-
is stopped loosely to son of gasolines the ratio of vapor to liquid volumes must be
avoid excessive evapo- set arbitrarily, and this adjustment is particularly convenient
ration. The funnel is with this apparatus. The ratio might be 1:9, corresponding
lowered again with the to a tank 90 per cent full. For fairly pure liquids, such as
I I stopcockopen, and then commercial acetone, ether, etc., a much higher ratio, five, to
shut off with volatile ten times the liquid volume, would be preferable. Results
liquid on both sides of agreeing well with literature values have been obtained for
the stopcock. Again ether, acetone, benzene, carbon tetrachloride, chloroform,
the funnel is raised and and pentane. The time for a test is 5 to 10 minutes; for
the new height of mercury (a) read as before. The difference check determinations, less than 1 minute.
HE specific gravity of semi-solid bituminous substances both ends and cut longitudinally in two equal parts, which
INCE the various methods for the determination of less degraded cellulose, their idea is the basis of the best
U. S. FORDST
PRODUCTS MADISON,WIS.
LABORATORY,
INCE the various methods for the determination of less degraded cellulose, their idea is the basis of the best
Ii Il
Stop C o c k C
81
nation treatment on wood, the crucible is water-jacketed
(this may be dispensed with in the case of pulps) in order to
I1 !kI
retard hydrolysis of the cellulose by the hydrochloric acid
formed. Oxidation of the cellulose is reduced to a minimum
by treating the washed chlorinated residue in the crucible
with dilute sulfurous acid solution immediately after chlorina-
5uref ktop Cock D
tion. The objections to using a calico or muslin pad in a l e v e l i n q Tube
Gooch crucible have been overcome, since the chlorinations
are carried out in a chamber consisting of a Jena (fritted-glass-
bottom) crucible. Chorination is expedited, as the chlorine
gas is not drawn through the sample by suction but forced
through under a slight hydrostatic pressure, thereby in-
creasing the speed of reaction. The moisture of the material
is evenly distributed through the entire sample, which gives a
Figure 1-Apparatus
u W a t e r Jacke f
S a t u r a f e d CB Cl, J o l u t i o n
for D e t e r m i n a t i o n of C h l o r i n e Absorp-
tion in Isolation of Cellulose
repeated until the fibers are a uniform white or show only a the findings of Mahood and Cable,2 analyses were made on
very faint tinge of pink upon addition of the sodium sulfite portions of the wood ground to three different degrees of
solution. The second and following treatments with chlorine fineness. A log having seventy annual growth rings was
should not require more than 1or 2 minutess each. quartered, air-dried at about 50 C., and cut into sawdust.
After all the lignin has been removed, the fibers are thor- This sawdust was ground to pass a 40-mesh standard sieve
oughly washed into the original alundum crucible (the fritted and the portion that remained on a 60-mesh sieve was placed
Jena glass crucible may be used instead of the alundum cru- in cloth bags of 75 grams capacity. These were extracted with
cible throughout the whole procedure) (porosity 98) with the alcohol-benzene mixture and designated sample 396.
approximately 500 cc. of hot distilled water and stirred oc- Some of this 40-60 mesh extracted sawdust was then reground
casionally with a pointed glass rod, then washed with 50 cc. until it passed a 60-mesh sieve and the part that was retained
of 95 per cent alcohol and finally with 50 cc. of ether. The on an 80-mesh sieve was designated as sample396B. The
tared crucible and its contents are dried for 2/2 hours or to material passing the 80-mesh and remaining on a 100-mesh
constant weight at 105 C. in an air oven, cooled in a desic- sieve was designated sample 396A.
cator over sulfuric acid, and weighed. I n Shorgers method the periods of chlorination were 30,
CHLORINECoNsuMmoN-The total volume of chlorine 20, 15, and 10 minutes, respectively; in both Ritters and
consumed in the isolation of the cellulose is obtained from the the authors methods the sawdust was subjected to four
sum of all the volumes used in the several chlorination treat- 3-minute chlorination periods. The data recorded under
ments, corrected individually to 0 C. and 760 mm. pressure, sample Val8represent white spruce wood, which was analyzed
according to the following formula: several years previous by the method of Schorger. It is
vt (P - a) added for comparative purposes only.
vo =
760 (1 + 0.003671) Application of Schorgers method gave higher yields of
cellulose from 80-100 mesh sawdust than were obtained from
where Vo = volume in cubic centimeters under standard
40-60 mesh material. The cellulose obtained from the finer
conditions; a = vapor pressure in mm. mercury of chlorine
ground material, however, was degraded to a greater extent
gas over calcium chloride solution saturated with chlorine
than that obtained from the coarser ground sawdust. This
gas a t some temperature, t, of chlorination; Vt = volume of is evidenced by the low yields of alpha-cellulose and also by
chlorine gas a t chlorination temperature, t; P = barometric the physical characteristics of the alpha-cellulose-namely,
pressure in mm. mercury. its gelatinous condition. These results compare well with
Note-Calcium chloride solution for the gas buret and the Hempel those of sample Va, which had been treated in the same
gas pipet: A solution saturated at room temperature with respect both to
calcium chloride and chlorine gas is prepared and filtered fpr use.
manner.
I n using the short chlorination method of Ritter, higher
Table I furnishes data for making the calculations accord- values for cellulose were also obtained with the 80-100 mesh
ing to the above formula. material than with the 40-60 mesh sawdust. However, by
Table I-Constants for Chlorine Consumption
the short chlorination methods the cellulose from the finer
* c. a 760 X (1 + 0.003871) ground material was not degraded as in Schorgers method;
24.6 17.3 as shown by the uniformly high values for alpha-cellulose.
26.0 18.0 These experiments confirm the work of Ritter-namely, that
28.0 19.4
27.0
28.0
20.8
22.0
the short chlorination period gives higher yields of alpha-
29.0 23.6 cellulose-indicating less degradation of the cellulose than is
30.0 26.0 caused by the longer periods of chlorination.
31.0 26.4
32.0 29.0 Chlorination by the crucible method described in this
paper gave approximately the same values for cellulose
The percentage of chlorine consumed may be obtained by regardless of the size of particle used in the experiments-
substitution of the proper values in the following formula: namely, 40-60,6040, and 80-100 mesh sawdust. I n addition,
Vo X 0.0031675 X 100 practically the same percentage of chlorine was consumed in
Per cent chlorine consumed = 0. D. weight of sample isolating the cellulose, although the samples of wood were
where the weight of 1 cc. of chlorine gas at standard tempera- ground t o 3 degrees of fineness.
ture and pressure = 0.0031675gram. The alpha-cellulose content is slightly higher in the cellulose
The percentage of chlorine forming hydrochloric acid in obtained from the coarser material. This indicates that the
the process is obtained by titration of the water used in wash- cellulose from the finer ground sawdust is hydrolyzed to a
ing the sample immediately after the several chlorination slightly greater extent than that from the coarser sawdust.
treatments. The final permanganate bleach of Schorgers method was
omitted in the new crucible method, as this procedure does
Results not appear to affect the yield of cellulose or the lignin content
To establish data relative to the reproducibility of results of the cellulose.
by the new crucible method for the estimation of cellulose and The lignin content of the cellulose samples was identical
to obtain comparative data showing the relationship of this regardless of whether the long chlorination periods of Schorger
method to the other methods from which i t was derived, sev- or the short chlorination periods of the new crucible method
eral series of experiments, results of which are recorded in were used. This confirms the work of Ritter,s in which
Table 11, were performed on a sample of sound white spruce comparison was made of the lignin content of cellulose ob-
wood, Picea glauca (Moench) Voss. tained in Schorgers apparatus by the long and short periods
The data show the comparative yields of cellulose obtained of chlorination. The 0.6 per cent of lignin, remaining in the
by three methods of analysis namely, Schorgers, Ritters, cellulose, is very difficult to completely remove with sodium
and the crucible method described herein. I n the first two sulfite even after the prolonged periods of chlorination. This
methods the wood was chlorinated in beakers placed in fact is further substantiated by Hauglg in his work on the
Schorgers chlorination apparatus. cellulose determination of cereal straws.
I n order to show the effect of the size of particle on the 18 Kress and others, U. S. Dept. Agr., Bull. 1298 (1926).
results obtained by the crucible method in comparison with 19 Haug, Uber die Natur der Cellulose aus Getreidestroh, Berlin, 1916.
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 43
Table XI-Comparative Analytical D a t a on Cellulose of Alcohol-Benzene Extracted S p r u c e Wood, O b t a i n e d by T h r e e M e t h o d s
(Results based on oven-dry weight of original wood)
LIGNIN LIGNIN
FINENESS AL- IN CONSUMPTION FINENESS AL- IN CONSUMPTION
OF PHA-CEL- OF CHLORINE OF PBA- C E L - OF CHLORINrt
SAM-DBTN. MA- CELLU- CELLU- LU- As SAM- DETN. MA- C E L L U - C E L L U - LU- As
PLE No. TBPIAL LOSB LOSE LOSE Total HCI RRMAXKS PLE No. TERIAL LOSE LOSE LOSE Total HCI REMARKS
Mesh % % % % % Mesh % % % % %
SCHORGBRS METHOD BRAYSb METHOD
Va . ., 60-80 58.537.1 .. .. *. U. S. De t Agr., 396 116 40-60 62.0 44.8 .. 37.1 .. Glass crucible
throughout
Bull. li96
396 105 40-60 68.0 43.5 .. .. *. Final
bleach
KMnO4
396B
Average
101 60-80
62.6 45.3
63.1 44.7 .. 36.2
36.2 21.7 Alundum crucible
396 106 40-60 59.2 4 3 . 5 0.57 .. .. Final
bleach
KMnO4
396B 102 60-80 62.7 44 4 ..
for weighing
37.0 2 1 . 3 Alundum crucible
3968 103 80-100 60.0 36.80 0.60 .. .. Final
bleach
KMnO4 396B 103 60-80 62.7 43 8 ..
for weighing.
3 5 . 7 21.5 Alundum crucible
for weighing
396A 104 80-100 60.2 37.20 0.57 .. *. Final
bleach
KMnO4 396B 104 60-80 62.9 43.9 .. 3 4 . 8 20.6 Alundum crucible
for weighing
RITTERS METHOD 396B 105 60-80 6 3 . 0 44.2 .. 3 4 . 9 21.0 Alundum crucible
396 107 40-60 61.2 45.0 .. .. *. Final
bleach
KMnOd
396B 106 60-80 63.1 44.0 ..
for weighing
3 5 . 5 Lost Alundum crucible
396 108 40-60 61.5 44.7 .. .. .. Final
hlearh
KMnO4
396B 107 60-80 62.7 ..
for weighing
0 . 6 0 3 5 . 8 21.9 Alundum crucible
396A
396A
105 80-100
106 80-100
62.5 44.3
6 2 . 3 44.3 .... .... ..
..
~
No final
No final
bleach
bleach 396B 108 60-80 62.7 .,
for weighing
0.45d 36.7 21.5 Alundum crucible
for weighing
BRAYSb METHOD 396B 109 60-80 62.5 ., 0 . 6 0 36.8 22.7 Alundum crucible
396 103 40-60 62.7 .. .. 37.2 .. Alundum crucible
for weighing Average 6 2 . 8 44.2 35.8 2 1 . 3
for weighing
396 104 40-60 62.6 46.1 .. .. .. Alundum crucible 396A 101 80-100 62.7 43.7 .. 36.4 . , Alundum crucible
for weighing
396 109 40-60 62.6 46.0 .. 34.6 .. for weighing
Alundum crucible 396A 102 SO-I00 61.8 43.2 .. 36.9 ,. Alundum crucible
for weighing
396 110 40-60 63.0 46.0 .. 34.5 .. for weighing
Alundum crucible 396A 107 80-100 62 7 43 7 .. 35.2 .. Glass crucible
throughout
396 111 40-60 6 2 . 5 44.7 .. 35.9 ..
for weighing
Alundum crucible
396A 108 80-100 62.0 44 1 .. 37.4 , . Glass crucible
HEMICAL engineering laboratories are not complete The usual laboratory does not have such facilities. The
C without fractionating equipment of some type. The
theory with derived operating curves is usually followed
by the experimental work. Without laboratory tests of the
continuous operation assures constant conditions, making
i t possible to obtain excellent data. The batch still, although
requiring much less liquor and storage space than the con-
theoretical curves the student may hesitate to accept the tinuous type, presents the difficulty of control due to the
establishment of the working principles. constantly changing concentration in the kettle and column,
The equipment is of either the batch or the continuous with the time lag involved. The frequent adjustment of
type. Continuous-type equipment requires a large supply the reflux ratio is exceedingly difficult for students. This
of raw material and storage space for the residue and distillate. paper describes a combination of the two systems in which
1 Received July 6, 1928. a small quantity of liquor is recirculated continuously, thus.
January 15, 1929 INDUSTRIAL AND ENGINEERING CHEMISTRY 43
Table XI-Comparative Analytical D a t a on Cellulose of Alcohol-Benzene Extracted S p r u c e Wood, O b t a i n e d by T h r e e M e t h o d s
(Results based on oven-dry weight of original wood)
LIGNIN LIGNIN
FINENESS AL- IN CONSUMPTION FINENESS AL- IN CONSUMPTION
OF PHA-CEL- OF CHLORINE OF PBA- C E L - OF CHLORINrt
SAM-DBTN. MA- CELLU- CELLU- LU- As SAM- DETN. MA- C E L L U - C E L L U - LU- As
PLE No. TBPIAL LOSB LOSE LOSE Total HCI RRMAXKS PLE No. TERIAL LOSE LOSE LOSE Total HCI REMARKS
Mesh % % % % % Mesh % % % % %
SCHORGBRS METHOD BRAYSb METHOD
Va . ., 60-80 58.537.1 .. .. *. U. S. De t Agr., 396 116 40-60 62.0 44.8 .. 37.1 .. Glass crucible
throughout
Bull. li96
396 105 40-60 68.0 43.5 .. .. *. Final
bleach
KMnO4
396B
Average
101 60-80
62.6 45.3
63.1 44.7 .. 36.2
36.2 21.7 Alundum crucible
396 106 40-60 59.2 4 3 . 5 0.57 .. .. Final
bleach
KMnO4
396B 102 60-80 62.7 44 4 ..
for weighing
37.0 2 1 . 3 Alundum crucible
3968 103 80-100 60.0 36.80 0.60 .. .. Final
bleach
KMnO4 396B 103 60-80 62.7 43 8 ..
for weighing.
3 5 . 7 21.5 Alundum crucible
for weighing
396A 104 80-100 60.2 37.20 0.57 .. *. Final
bleach
KMnO4 396B 104 60-80 62.9 43.9 .. 3 4 . 8 20.6 Alundum crucible
for weighing
RITTERS METHOD 396B 105 60-80 6 3 . 0 44.2 .. 3 4 . 9 21.0 Alundum crucible
396 107 40-60 61.2 45.0 .. .. *. Final
bleach
KMnOd
396B 106 60-80 63.1 44.0 ..
for weighing
3 5 . 5 Lost Alundum crucible
396 108 40-60 61.5 44.7 .. .. .. Final
hlearh
KMnO4
396B 107 60-80 62.7 ..
for weighing
0 . 6 0 3 5 . 8 21.9 Alundum crucible
396A
396A
105 80-100
106 80-100
62.5 44.3
6 2 . 3 44.3 .... .... ..
..
~
No final
No final
bleach
bleach 396B 108 60-80 62.7 .,
for weighing
0.45d 36.7 21.5 Alundum crucible
for weighing
BRAYSb METHOD 396B 109 60-80 62.5 ., 0 . 6 0 36.8 22.7 Alundum crucible
396 103 40-60 62.7 .. .. 37.2 .. Alundum crucible
for weighing Average 6 2 . 8 44.2 35.8 2 1 . 3
for weighing
396 104 40-60 62.6 46.1 .. .. .. Alundum crucible 396A 101 80-100 62.7 43.7 .. 36.4 . , Alundum crucible
for weighing
396 109 40-60 62.6 46.0 .. 34.6 .. for weighing
Alundum crucible 396A 102 SO-I00 61.8 43.2 .. 36.9 ., Alundum crucible
for weighing
396 110 40-60 63.0 46.0 .. 34.5 .. for weighing
Alundum crucible 396A 107 80-100 62 7 43 7 .. 35.2 .. Glass crucible
throughout
396 111 40-60 6 2 . 5 44.7 .. 35.9 ..
for weighing
Alundum crucible
396A 108 80-100 62.0 44 1 .. 37.4 , . Glass crucible
HEMICAL engineering laboratories are not complete The usual laboratory does not have such facilities. The
C without fractionating equipment of some type. The
theory with derived operating curves is usually followed
by the experimental work. Without laboratory tests of the
continuous operation assures constant conditions, making
i t possible to obtain excellent data. The batch still, although
requiring much less liquor and storage space than the con-
theoretical curves the student may hesitate to accept the tinuous type, presents the difficulty of control due to the
establishment of the working principles. constantly changing concentration in the kettle and column,
The equipment is of either the batch or the continuous with the time lag involved. The frequent adjustment of
type. Continuous-type equipment requires a large supply the reflux ratio is exceedingly difficult for students. This
of raw material and storage space for the residue and distillate. paper describes a combination of the two systems in which
1 Received July 6, 1928. a small quantity of liquor is recirculated continuously, thus.
44 ANALYTICAL EDITION Vol. 1, No. 1
giving constant conditions that are applicable to the batch kettle indicates kettle concentrations. The piping from con-
and continuous types. denser to kettle converts the batch process into a continuous
Figure 1 shows the complete distillation equipment. The operation. The condensate from the condenser passes
30-gallon still, plate column, and condenser are of the usual through two Venturi meters, one for the reflux and the other
for the net output. With identical meters, the square root
of the quotient of the manometer readings gives ihe reflux
ratio, the liquids metered being of similar composition. The
Venturi manometers are of the 2-liquid multiplying type.
The reflux ratio is regulated by the valve A . The open vent
a t B eliminates syphoning, thereby insuring a uniform flow
through either meter.
The operation is as follows: Fill the kettle to the 20-
gallon mark with a binary mixture and bring to the distilla-
tion temperature. Build up the condensate composition
by returning all the condensate as reflux through operation
of valve A . When the desired condensate composition is
reached permit a portion to flow continuously through the
sight gravity jar into the kettle. When constant conditions
are assured, take readings of the Venturi manometers, kettle
thermometer, and condensate hydrometer. The conditions
evidently represent those obtainable at some moment during
a batch run.
By directing a fraction of the condensate through the cock
D into a storage vessel, the kettle concentration can be de-
creased and a new set of conditions established. From a
series of such findings, the operation of the batch type can
be represented by a plot of reflux ratio us. kettle concentra-
tion. For easy reference the same relationship can be ex-
pressed with the square root of the quotient of the manometer
readings plotted against kettle temperatures.
Curves for continuous operation, effect of boiling rate,
Figure 1 d a t e and thermal efficiencies as functions of amrotxiate
I. *
variables can be obtained in a similar way. The equipment
design. Thermometers placed in wells in the hand hole described is operated with very satisfactory results, in the
covers of each plate are useful in observing temperature Chemical Engineering Laboratory of the C'niversity of Wash-
changes. The thermometer placed in the vapor of the ington.
N A number of industrial processes alkaline solutions are During the past ten years pH control by the colorimetric
N A number of industrial processes alkaline solutions are During the past ten years pH control by the colorimetric
Per cent #H 6H
Indigo carmine Indigo disulfonate 0.1 Water b. 11.6-14.0 J. b. 11.6-13.2 J . Fading; unstable reagent
........
Malachite green
Indigo tetrasulfonate
Tetra-methyl di-$-aminotriphenyl-
0.1 Water .......... b. 1 1 . 0 - 1 3 . 0 ~ . Fading; unstable reagent
Water blue
carbinole
Di- and trisulfonic acids of tri-
0.1 Water b. 11.5-14.0 f. ......... Fading
phenyl-+ and diphenylrosaniline 0.1 Water .......... pu. 10.0-12.0 hr. Fading
Alkali blue Monosulfonic acids of diphenyl-
rosaniline and triphenyl-$-ros-
aniline 0.2 Alcohol V. 9 . 4 - 1 4 . 0 ~ . h. 11.4-14.0~. Fading; unstable reagent
Methyl blue Di- and trisulfonic acids of tri-
phenylrosaniline 0.1 Water b. 1 0 . 0 - 1 3 . 0 ~ . b. 1 0 . 0 - 1 2 . 4 ~ . Fading
Fuchsin, acid Di- and trisulfonic acids of rosani-
line and #-rosaniline 0.1 Water r. 12.0-14.0 f. ......... Fading
Fuchsin, basic Rosaniline and #-rosaniline 0.1 Water .......... ......... Fading. unstable reagent
Crystal violet
Alizarin blue S
Hexamethyl-$-rosaniline
NaHSOs compound of dihydroxy-
0.05 Water .......... ......... Fading: unstable reagent
anthroquinone-@-quinoline 0.1 Water .......... ......... Fading; variable colors;
unstable reagent
Alizarin 1,2-Dihydroxyanthroquinone 0.04 Water v. 10.1-12.1 p. ......... Variable colors; un-
stable reagent
Nile blue A Diethyldiaminonaphthophen-
azoxonium sulfate 0.04 Water b. 10.2-13.0 P. b. 9.6-10.4 p. Fading; fluorescence
Purpurin 1,2,4-Trihydroxyanthroquinone
..........................
0.1 Water .......... ......... Fading; variable colors
Poirriers blue 0.1 Water b. 11.0-13.0 r. pu. 10.0-11.5 br. Fading
........ Nitrothymolsulfonphthalein 0.04 Water v. 9.2-11.5y. v. 9.2-11.5 y. Fading
........ Nitro henolsulfonphthalein 0.04 Water .......... r. 9 . 5 - 1 1 . 2 ~ . Fading
........ 1 3 5 - Grinitrobenzene
0.2 Alcohol c. 12.0-14.0 0. c. 11.4-13.0 0. Fading
T. N. T. 2:4:6-Trinitrotoluene 0.1 IAlcohol 6C
Nitramine, tetryl 2,4,6-Trinitrophenyl methylnitro-
L Water 40 p. 11.5-14.0 0. ......... Color development slow
amine 0.1 I Alcohol 6C
(Water 40 c. 10.8-13.0 br. c. 10.8-13.0 br. Fading
........ Dinitrohydroquinone diacetate 0.1 f Alcohol 60
1 Water 40 .......... ......... Color development slow
Methyl green
Eosin-methylene blue
Methylated crystal violet
..........................
0.1
0.1
Water
Water
..........
.......... b.
.........
13.0-13.8~.
Fading
Pptn: unstable reagent
Safranine 0 Di- and trimethylphenosafranine 0.1 Water .......... ......... Preciiitation
Thymolphthalein 0.1 Alcohol C. 9.3-10.5 b. C. 9.2-10.4 b. Precipitation
........ m-Cresolphthalein 0.1 Alcohol .......... c. 9.2-10.4v. Precipitation
Cresyl blue .......................... 0.05 Alcohol .......... b. 10.8-11.4 p. Precioitatlon
Fast red A 5-Sulfonaphthalene-azo-@-naphthol 0.1 Water 10.5-12.1 p. 10.4-12.0 0. Chanie not pronounced;
unstable reagent
Fast red B - -
a-Naphthalene azo @-naphthol.
3,6-disulfonic acid 0.1 Water p. 10.5-12.5 0. p. 10.5-12.5 0. Change not pronounced;
unstable reagent
Orange G Benzene azo-8-naphthol-7-disul-
fonic acid 0.1 Water y. 11.5-14.0 p. 0. 11.2-12.4 r. Change not pronounced
Benzopurpurin B - -
Ditolyldisazo his @ naphthyl-
amine-6-sulfonic acid 0.05 Alcohol y. 12.0-14.0 r. ......... Changes not pronounced
Eosin B N
Croceine scarlet 7 B
Dibromodinitrofluorescein
..........................
0.1
0.1
Water
Water
p. 10.5-14.0 y.
.......... .........
.........
No change; fluorescent
No change
Alpha-naphthylamine R salt
Carminic acid
..........................
..........................
0.05
0.1
Water
Alcohol
..........
..........
m. 10.8-11.6 p.
v. 11.0-12.0 p.
Change not pronounced
Change not pronounced
Phenacetolin .......................... 0.1 Alcohol r. 10.0-13.0 C. r. 10.0-13.0 y. Unstable reagent
Orange 11 $-Benzenesulfonic acid azo-@-
Clayton yellow
naphthol
..........................
0.05
0.1
Water
Water
..........
..........
0.
y.
10.6-12.0r.o.
12.8-14.0 0.
Change not pronounced
..................
Alizarin yellow GG m-Nitrobenzeneazosalicylic acid 0.1 Water c. 10.0-12.0 y. 1. 10.0-12.0 y. ..................
Alizarin yellow R 9-Nitrobenzeneazosalicylic acid 0.1 Water y. 10.0-12.1 y. y. 10.0-12.0 0. ..................
Tropaolin 0 p-Sulfobenzeneazoresorcin 0.05 Water 7.11.1-12.70. y. 11.0-13.00. ..................
Azo blue Ditolyldisazo-bis-a-naphthol-4-5111-
fonic acid 0.04 Water v. 10.5-11.5 p. V. 10.0-11.0 p. .................
........ or-Kauhthol benzein 0.1 Alcohol v. 8 . 5 - 9 . 8 -~
~ . V. 9.0-10.6c. .................
Abbreviations are as follows: b blue, br brown. c cc rless; f ,
v, violet; and y, yellow. I. C. T. refers to Irherdational Critica rables, F . I., pp. 84-88.
cured from different sources. A few dyes not obtainable on Above pH 12.6 it was necessary to use reference solutions of
the market were prepared in the writers laboratory. sodium hydroxide. This required the use of variable molal
Indicator solutions were prepared by dissolving weighed concentrations, which is objectionable in the accurate stand-
amounts of the dyes in ethyl alcohol or water, adjusting the ardization of an indicator, although not in a general investi-
concentration to give the most satisfactory intensity when gation such as this. I n all cases the pH of the reference
0.5 cc. of the indicator solution was added to 10 cc. of the solu- solution was determined by means of a hydrogen electrode a t
tion to be tested. All dye solutions were kept in glass-stop- 25 * 1OC.
pered bottles for later observat,ion as to their stability. Each indicator was examined by mixing 0.5 cc. of the indi-
A series of buffer mixtures was prepared extending from cator solution with 10 cc. of each of the buffer mixtures in a
pH 10.0 to pH 14.0 in intervals of about 0.2 pH. The pH tube 14 to 16 mm. in diameter and observing the appearance of
of each solution was carefully determined by means of a hy- this mixture by transmitted light, all tests being made at
drogen electrode, and all solutions were used a t once to guard 25 * 1 C. By means of this test the various dyes were
against any change. Several types of these mixtures were evaluated accordingly as they fulfilled the qualifications of a
employed, since it was found that some dyes gave different satisfactory indicator for practical use. Such an indicator
color changes in mixtures having different compositions, even should give well-defined color changes over a relatively
when the pH was the same. This probably explains why short range; it should not be unduly affected by substances
different investigators have reported different color changes other than hydrogen or hydroxyl ions; it should immediately
for some dyes. The following buffer mixtures were used: give colors that are stable for a reasonable length of time, so
sodium carbonate-hydrochloric acid (pH 10.2 to 11.4),2 that there will be no errors due to changes taking place during
glycocoll-sodium chloride-sodium hydroxide (pH 10.0 to the determination; its solution should be a stable reagent.
12.6),3 and borate-sodium hydroxide (pH 10.0 to ~ z Z ) . ~ An indicator that fades rapidly, precipitates, undergoes change
2 Kolthoff, Indicators, p. 146 (1926).
in color due to the presence of salts, colloidal material, etc.,
* Sorensen and Walbum, International Critical Tables, Vol. I, p. 84 attains a stable color slowly, or is unstable as a reagent, is sure
(1926). to be objectionable in industrial control.
46 ANALYTICAL EDITION Vol. 1, No. 1
Discussion of Results centration of dye is purposely lowered to avoid precipita-
tion, the color is too faint except in a layer too deep for a
The results of these tests are shown in the accompanying practical test.
table. Some spaces are left unfilled either because no data With dyes which give fluorescence the color by reflected
were available or because the results were of such a nature light often differs radically from that by transmitted light.
that they could not be recorded without comment. The Eosin BN gave no marked color changes between pH 10.0 and
latter cases are discussed in the text. 14.0 when observed by transmitted light. Fluorescence was
I n those instances in which no pH range has been given, marked throughout the range and the tubes appeared yellow
either no change in color occurred or the change was appar- by reflected light. The color change for this dye is given as
ently due to decomposition or precipitation of the dye. For pink 10.5-14.0 yellow in the International Critical Tables.
example, up to pH 11.8 the blue color of malachite green was Certain of the dyes developed a stable color slowly. Dini-
fairly stable, but above this fading was almost instantaneous. trohydroquinone diacetate and trinitrotoluene are examples.
Such a compound, where the change is due to progressive When the former was added to buffer mixtures ranging from
fading, can hardly be classed as a pH indicator. Crystal pH 10.0 to 12.0, all the tubes were yellow a t first. I n the
violet, methyl green, and the fuchsins acted in a similar man- lower range this yellow color gradually deepened, while a t
ner. pH 11.0 and above, it slowly changed to a deep purpIe.
A peculiar phenomenon was noted in the case of alizarin and Although none of the dyes examined are as satisfactory as are
alizarin blue S. Clark' and Balch6 give conflicting color the sulfonphthaleins of Clark and Lubs,Band Cohen' at lower
changes for these compounds in the higher range. Clark, ranges, about five can be used for hydrogen-ion work. These
from data culled from the literature, reports alizarin (violet are alpha-naphthol benzein, Clayton yellow, alizarin yellow
10.1-12.1 purple) and alizarin blue S (green 11.0-13.0 blue), GG, alizarin yellow R, tropaolin 0, and azo blue. Objections
whereas experimental results reported by Balch are alizarin can be raised against each of these, however. The majority
(wine buff 9.8-11.4 deep violet) and alizarin blue S (brown give color changes that are not very pronounced, from yellow
9.5-12.0 blue green). The present writers found that such to deep yellow or orange. Alizarin yellow R has been found
conflicting results were probably due to the nature of the to give misleading lemon-yellow shades in solutions containing
buffer mixture used. With carbonate mixtures they were lime, although it is fairly satisfactory otherwise. Tropaolin
able to check the results given by Clark for alizarin, except 0 gives quite marked readings, although it has been reported
that the range was 10.8-12.0. With alizarin blue S the pres- to have a rather large salt error.2 Azo blue gives a good color
ent writers found blue 11.0-12.4 green, a reversal of the colors change in the pH range 10.0 to 11.0, but it is not entirely
given in Clark's table. However, when borate buffer mix- satisfactory much above or below this. Alpha-naphthol ben-
tures were employed, the writers' results checked with those zein gives fair color changes, but the dye is low in tinctorial
given by Balch. The color changes for purpurin also varied power.
with the buffer mixture. I n an investigation of this nature the matter of purity of
Precipitation is another source of error in determining the the dyes can always be raised. Since no spectrophotometric
color change, Buffer mixtures containing safranine a t first examination of the dyes was made, no definite proof as to
gave a red color which gradually changed to violet, due to their purity can be given. I n general, the investigation is
precipitation, which was quite pronounced a t high alkalini- an examination of dyes that have been recommended and in
ties. This may account for the color change, red to violet, many cases are being sold for the purpose of pH studies over
which is sometimes given for this dye.3 With thymolphthal- the range 10.0 to 14.0. As such, most of these dyes are de-
ein and m-cresolphthalein, precipitation is usually sufficient cidedly unsatisfactory.
to render these dyes decidedly unsatisfactory. It often causes Since this investigation has yielded no indicators of out-
a bluish cast, even in solutions having a pH below 9.0, which standing excellence, attempts are now being made to syn-
may be taken for the color of the indicator. When the con- thesize new dyes to cover the higher pH ranges.
4 Clark, "The Determination of Hydrogen Ions," p. 91 (1922). 6 Clark and Lubs, J . B a d . , 2, 1, 109, 191 (1917).
8 Balch, Sugar, 27, 587 (1925). 7 Cohen, U.S.Pub. Health Rcpts. 41, 3051 (1926).
H E rate of condensation of steam is largely affected water is indicated, and the space above is filled with the
H E rate of condensation of steam is largely affected water is indicated, and the space above is filled with the
mercury could be drawn into the instrument is substantially a gas thermometer, and the
the system as desired. The calibration curve of pressure against temperature is nearly
apparatus finally contained a straight line. This line gives the largest manometer read-
an amount of air in the bulb ing possible at a given temperature, and an addition of air
exerting a pressure of about to the steam space of the boiler reduces this reading. At a
400 mm. absolute, enough given temperature, therefore, the pressure read on the manom-
water in both sides to fill the eter is subtracted from that read on the calibration curve
tube and keep a layer in the and the difference is directly the partial pressure of air in
bulb, and sufficient mercury the boiler.
to indicate the pressures to The accuracy of the method is limited by the precision
be read. with which the manometer may be read, since Thompsons
If, for simplicity in ex- and writers of more recent books on thermodynamics have
planation, the hydrostatic shown that the vapor pressure of water is not affected appre-
pressure of water is neg- ciably by as much as several atmospheres pressure of an
lected in comparison to that indifferent gas. Readings to 0.5 mm. on an ordinary scale
of mercury, the operation of give an accuracy of 1 part in 2000 when the steam is slightly
the manometer is as follows: above atmospheric pressure.
At thermal equilibrium the This instrument combines the characteristics of a gas
partial pressure of the steam thermometer with those of a vapor-pressure thermometer,
in the bulb is the same as and other uses with other vapors and gases will probably be
that in the boiler and bal- found for it. If the bulb is filled with a liquid in equilibrium
ances or cancels it on the with its vapor in a pure state, still other uses are apparent,
manometer. Thus t h e such as the direct determination of the partial pressure of one
manometer reads directly liquid in a binary solution or the determination of the lower-
the difference of p r e s s u r e ing of the vapor pressure of water by the addition of an
Air Manometer a n d Connections between the air in the bulb electrolyte.
and that of the air in the * Thompson, Applications of Dynamics to Physics and Chemistry,
boiler. When no air is present in the boiler steam space, Macmillan, 1888.
mercury could be drawn into the instrument is substantially a gas thermometer, and the
the system as desired. The calibration curve of pressure against temperature is nearly
apparatus finally contained a straight line. This line gives the largest manometer read-
an amount of air in the bulb ing possible at a given temperature, and an addition of air
exerting a pressure of about to the steam space of the boiler reduces this reading. At a
400 mm. absolute, enough given temperature, therefore, the pressure read on the manom-
water in both sides to fill the eter is subtracted from that read on the calibration curve
tube and keep a layer in the and the difference is directly the partial pressure of air in
bulb, and sufficient mercury the boiler.
to indicate the pressures to The accuracy of the method is limited by the precision
be read. with which the manometer may be read, since Thompsons
If, for simplicity in ex- and writers of more recent books on thermodynamics have
planation, the hydrostatic shown that the vapor pressure of water is not affected appre-
pressure of water is neg- ciably by as much as several atmospheres pressure of an
lected in comparison to that indifferent gas. Readings to 0.5 mm. on an ordinary scale
of mercury, the operation of give an accuracy of 1 part in 2000 when the steam is slightly
the manometer is as follows: above atmospheric pressure.
At thermal equilibrium the This instrument combines the characteristics of a gas
partial pressure of the steam thermometer with those of a vapor-pressure thermometer,
in the bulb is the same as and other uses with other vapors and gases will probably be
that in the boiler and bal- found for it. If the bulb is filled with a liquid in equilibrium
ances or cancels it on the with its vapor in a pure state, still other uses are apparent,
manometer. Thus the such as the direct determination of the partial pressure of one
manometer reads directly liquid in a binary solution or the determination of the lower-
the difference of p r e s s u r e ing of the vapor pressure of water by the addition of an
Air Manometer a n d Connections between the air in the bulb electrolyte.
and that of the air in the * Thompson, Applications of Dynamics to Physics and Chemistry,
boiler. When no air is present in the boiler steam space, Macmillan, 1888.
R
1 '61 I I I 1 I I 1
700
I I 1 I
8 /, A 20 24 32
T/ME IN HOURS
BEAT-
ING
TIME
I
MULLENSTRENGTH
Pulp A Pulp
FACTOR
(100 X kg. per sq. cm. per kg.
ream weinht)
Ratio 1 Pulp A
ENDURANCB
FOLDING
Pulp B Ratio
II TBNSILE
Pulp A
STRENGTH
Pulp B Ratio
II RATE
Pulp A
OF CHANGE OF
FREBNESS
Pulp B Ratio
sideration of these facts and Figures 1 to 4, in which the data F = 706 - 108.3
are plotted, it is evident that the freeness-time data alone and the first' derivative is given by
are suited to any considerable mathematical treatment,
largely because such data are obtained by working with wet,
almost fluid material under conditions where sampling is
easy and accurate. The bursting strength data are next best, Table I1 gives a comparison of values for the two pulps
but cannot compare in accuracy with the freeness values and, together with the ratio of the A to the B value a t various
except for only part of the range, the bursting strength bears times of beating.
no very simple relation to the beating time. Conclusions
It has been suggested3 that for the same pulp the rate of
change of freeness with time affords a satisfactory means of From a consideration of the plots and tabular data, it is
comparing the performance of two beaters. Conversely, evident that pulp A is by far the superior pulp, that for pulp
using the same beater, the rates of change of freeness with A the maximum fold is developed after beating 6 hours, that
time should afford a valuable comparison of the ease with both less and more beating than that accomplished in 6 hours
which two pulps may be beaten. Figure 1is a plot of freeness result in considerably decreased folding endurance, that
against time for the two pulps, A and B, considered here, and shrinkage for both pulps becomes linear with time after 4
the slopes of these curves are the rates of change of freeness hours' beating, and that the bursting strength increases very
with respect to time. These slopes cannot be determined little after 10 hours' beating. There is no reason why such
graphically with accuracy, but it is evident that they are comparative data could not be duplicated by the use of any
given by the first derivatives of freeness with respect to time beater. However, instead of beating to the time intervals of
and are readily found once the equations connecting freeness this test, i t would be necessary to beat to the freeness values
and time are known. Fairly simple exponential equations reported a t these time intervals.
were found to fit these curves very closely, as indicated by The ratio of Mullen strength factors as shown in Table I1
a comparison of actual and calculated freeness values listed varies from 2.42 to 1.85, or 24 per cent, the ratios of folding
in Table I. For pulp A the equation is endurances from 250 to 30, or 88 per cent; the ratios of
F = 676 - 118 @Jo tensile strengths from 2.5 to 1.5, or 40 per cent; the ratios
of rates of change of freeness from 1.50 to 1.77,or 18 per cent,
and the first derivative is which is the most constant. This, together with the ease,
d_F = -59 e-o.60 accuracy, and rapidity of the freeness tests and their sus-
de ceptibility to mathematical treatment, leads to the suggestion
For pulp B the equation is that rate of change of freeness be used as a preliminary means
8 Davis, Pulp PaDer Mag. Can., 24, 987 (1926). of comparing pulps.
BEAT-
ING
TIME
I
MULLENSTRENGTH
Pulp A Pulp
FACTOR
(100 X kg. per sq. cm. per kg.
ream weinht)
Ratio 1 Pulp A
ENDURANCB
FOLDING
Pulp B Ratio
II TBNSILE
Pulp A
STRENGTH
Pulp B Ratio
II RATE
Pulp A
OF CHANGE OF
FREBNESS
Pulp B Ratio
sideration of these facts and Figures 1 to 4, in which the data F = 706 108.3 -
are plotted, it is evident that the freeness-time data alone and the first' derivative is given by
are suited to any considerable mathematical treatment,
largely because such data are obtained by working with wet,
almost fluid material under conditions where sampling is
easy and accurate. The bursting strength data are next best, Table I1 gives a comparison of values for the two pulps
but cannot compare in accuracy with the freeness values and, together with the ratio of the A to the B value a t various
except for only part of the range, the bursting strength bears times of beating.
no very simple relation to the beating time. Conclusions
It has been suggested3 that for the same pulp the rate of
change of freeness with time affords a satisfactory means of From a consideration of the plots and tabular data, it is
comparing the performance of two beaters. Conversely, evident that pulp A is by far the superior pulp, that for pulp
using the same beater, the rates of change of freeness with A the maximum fold is developed after beating 6 hours, that
time should afford a valuable comparison of the ease with both less and more beating than that accomplished in 6 hours
which two pulps may be beaten. Figure 1is a plot of freeness result in considerably decreased folding endurance, that
against time for the two pulps, A and B, considered here, and shrinkage for both pulps becomes linear with time after 4
the slopes of these curves are the rates of change of freeness hours' beating, and that the bursting strength increases very
with respect to time. These slopes cannot be determined little after 10 hours' beating. There is no reason why such
graphically with accuracy, but it is evident that they are comparative data could not be duplicated by the use of any
given by the first derivatives of freeness with respect to time beater. However, instead of beating to the time intervals of
and are readily found once the equations connecting freeness this test, i t would be necessary to beat to the freeness values
and time are known. Fairly simple exponential equations reported a t these time intervals.
were found to fit these curves very closely, as indicated by The ratio of Mullen strength factors as shown in Table I1
a comparison of actual and calculated freeness values listed varies from 2.42 to 1.85, or 24 per cent, the ratios of folding
in Table I. For pulp A the equation is endurances from 250 t o 30, or 88 per cent; the ratios of
F = 676 -
118 @Jo tensile strengths from 2.5 to 1.5, or 40 per cent; the ratios
of rates of change of freeness from 1.50 t o 1.77,or 18 per cent,
and the first derivative is
which is the most constant. This, together with the ease,
d_F = -59 e-o.60 accuracy, and rapidity of the freeness tests and their sus-
de ceptibility to mathematical treatment, leads to the suggestion
For pulp B the equation is that rate of change of freeness be used as a preliminary means
8 Davis, Pulp PaDer Mag. Can., 24, 987 (1926). of comparing pulps.
I
the powdered cupric hydroxide.
Sampling
The cellulose used in making up a solution should be selected
by careful sampling, as different portions of a bale may differ
very much. There are teasing machines on the market which
might be used for mixing cellulose, but many would not find it
convenient to have one. A good way is to take samples from
many parts of the bale, or bales, to be tested, tear them into
fine shreds, and mix them carefully. This may be done by throw-
ing the pieces on a table as they are shredded, then quartering
and repeating until a sample of suitable size has been obtained.
This gives a more representative sample and more uniform re-
sults than if the sample were selected from only one place.
Careful sampling is essential if consistent results are to be ob-
tained. ll
Concentration of Cellulose
The weight of cotton to be used in making up solutions should
be carefully considered. The falling-sphere viscometer later de-
scribed can be used with substances of from 30 to 300,000 centi-
poises, although its best range is from 100 to 10,000 centipoises.
Celluloses vary so widely that no concentration can be chosen
which will be sure to give viscosities within these limits. There-
fore, some laboratories use 2.5 grams of cotton per 100 cc. of
solution for ordinary cotton, 5.0 grams for very low-viscosity
cottons, and 1.0 gram for high-viscosity cottons. The last two
viscosities are then changed to the viscosities of solutions con-
taining 2.5 grams of cotton by the application of Joyner's law.2 c
Determination of Alpha-Cellulose'
Report of Work of Sub-committee 2 of the Division of Cellulose Chemistry of the
American Chemical Society
George J. Ritter. Chairman
Method I
% % -1 % % I % % I % % 1 % %
99.50 $0.07 91.64 -0.86 88.66 -0.17 83.21 -0.30 82.97 $0.61
99.62 $0.19 95.21 $2.70 87.71 $0.88 85.90 4-2.39 84.68 +2.32
99.52 $0.09 91.36 -1.14 86.37 -0.46 83.02 -0.49 86.63 $0.27
99.17 -0.26 91.70 -0.80 86.79 -0.04 83.57 $0.06 82.42 $0.06
99.36 -0.07 92.36 -0.14 86.33 -0.50 82.24 -1.27 80.14 -2.22
99.09 -0.34 91.59 -0.91 86.44 -0.39 83.21 -0.30 81.96 -0.40
99.60 $0.17 93.56 $1.06 87.05 $0.22 82.92 -0.59 82.79 $0.43
99.60 $0.17 92.60 $0.10 87.34 $0.51 84.05 $0.54 81.30 -1.06
- -
Average 99.43 0.17 I 92.50 0.96 1 86.83 0.39 I 83.51 0.75 I 82.36 0.92
Method I1
A 99,26 -0.08 92.38 -0.13 86.63 -0.22 83.85 $0.47 82.74 +0.51
B 99.63 $0.29 95.09 12.58 88.34 $1.49 85.63 12.25 83.95 $1.72
C 99.42 $0.08 91.95 -0.56 86.82 -0.03 82.15 -1.23 81.68 -0.55
D 99.30 -0.04 93.00 10.49 86.57 -0.18 83.07 -0.31 81.85 -0.38
E 99.29 -0.05 91.35 -1.16 86.41 -0.44 82.54 -0.84 80.25 -1.98
F 99.20 -0.14 91.67 -0.84 86.28 -0.57 83.12 -0.26 81.85 -0.38
G 99.26 -0.08 92.14 -0.37 86.92 $0.07 83.34 -0.04 83.32 $1.09
Average 99.34 0.11 92.51 0.88 86.85 0.43 83.38 0.77 82.23 0.94
Method I11
COTTON
40 I COTTON
50 I PULP 10 I PULP 20 I PULP30
99.84 $0.52 93.46 $0.52 87.35 $0.34 84.68 $0.49 84.59 $0.96
99.35 $0.03 92.56 -0.38 86.71 -0.30 84.31 10.12 83.36 -0.27
99.41 $0.09 94.02 11.08 87.13 $0.12 85.55 11.36 83.65 $0.02
99.78 $0.46 91.86 -1.08 86.50 -0.51 84.01 -0.18 83.34 -0.29
99.31 -0.01 92.80 -0.14 86.95 -0.06 83.53 -0.66 83.28 -0.35
99.50 $0.18 92.22 -0.72 86.32 -0.69 82.63 -1.56 82.19 -1.44
99.60 $0.28 93.60 $0.66 87.47 $0.46 84.40 $0.21 84.70 $1.07
97.87 -1.45 93.04 +0.10 87.70 -0.69 84.46 +0.27 83.98 +0.35
Average 99.32 0.37 I 92.94 '0.58 I 87.01 0.39 1 84.19 0.60 I 83.63 0.59
Method IV
COTTON
C-100 ] COTTON C-200 1 PULPT-500 I PULP"-600
I
99.30 $0.06 93.06 $1.59 86.35 -0.01 82.56 -0.26
99.00 -0.24 91.33 -0.14 86.26 -0.10 83.10 $0.28
99.47 $0.23 91.11 -0.36 86.33 -0.03 82.56 -0.26
99.37 +0.13 91.07 -0.40 85.78 -0.58 81.58 -1.24
99.44 +0.20 93.50 $2.03 87.03 4-0.67 84.22 $1.40
99.27 $0.03 90.06 -1.41 86.53 +O. 17 83.82 i-1.00
99.32 $0.08 91.30 -0.17 86.58 $0.22 82.37 +0.05
98.84 -0.40 90.78 -0.69 86.00 -0.36 81.67 -1.15
99.19 -0.05 91.04 -0.43 86.36 0.00 83.02 10.20
Average 99.24 0.15 1 91.47 0.80 I 86.36 0.23 I 82.82 0.64 I
54 ANALYTICAL EDITION Vol. 1, No. 1
to room temperature, and its strength determined by titration with a mination should be weighed out at the same time that the sam-
standard acid. I t is then diluted to give a solution containing 17.5 per ples for the alpha-cellulose determination are taken.
cent by weight of sodium hydroxide (17.5 grams of sodium hydroxide in
100 grams of the solution). NOTES ON METHODIV-(1) Preparation of Sodium Hydroxide Solution.
(2) Time Interval for Steps i n Procedure. An interval of 10 minutes The sodium hydroxide solution is prepared by dissolving sticks of alkali in
should elapse between the times of starting the alkali treatment of the in- a n equal weight of water, and allowing i t t o stand 10 days for the sodium
dividual triplicate samples. Such a procedure allows time before proceeding carbonate and impurities t o settle. The clear, supernatant liquid is de-
t o the next sample in the triplicate determination for the various steps such canted and diluted with carbon dioxide-free water until its density at 15' C.
a s impregnating, macerating, filtering, and partial washing. is 1.197. Such a solution contains 17.5 * 0.1 grams of sodium hydroxide
If 2 minutes are allowed between the addition of the alkali and the be- per 100 grams of solution.
ginning of maceration, i t will be found t h a t sulfite pulp swells t o five or six (2) Time Interval for Steps in the Procedure. An interval of 15 minutes
times its original volume, and the fibers can be much more easily macerated, should elapse between the times of starting the alkali treatment of the
The remaining 8-minute interval has been found ample time for the macer- individual triplicate samples. Such a procedure allows time before pro-
ation of most pulps. With such a scheme, for example, the alkali would be ceeding t o the next sample in the triplicate determination for the various
added to the samples a t 9:00 o'clock, 9:10, and 9:20, respectively. Filtering steps in the procedure, such a s impregnation, maceration, filtration, and
would begin a t 9:45 o'clock, 9:55, and 10:05, respectively. partial washing.
(3) Filtering Medium. A coarse Jena glass filter crucible may be used If 5 minutes are allowed between the addition of a part of the alkali and
in place of the Gooch crucible specified in the directions. the beginning of maceration, it will be found t h a t sulfite pulp swells t o five or
(4) Testingfor Acidity. To test the acidity of the wash water on litmus six times its original volume, and the fibers can be much more easily macer-
paper, the last few drops removed after the addition of water should be ated. The remaining 10-minute interval, during which the remaining 40-cc.
used. portion of the alkali is added, making a total of 75 cc., has been found ample
(51 Drying lo Conslant Weight. The constant weight obtained in a time for the maceration of pulps. With such a scheme, for example, one
minimum time should be used-that is, the first constant consecutive weights would add the alkali to the samples at 9:00 o'clock, 9:15, and 9:30, re-
at hour intervals after approximately 4 hours of drying. It is reported spectively. Filtering would begin at 9:45 o'clock, 10:00, and 10: 15, respec-
t h a t cellulose dried 12 t o 16 hours increases in weight. tively.
(3) Filtering Medium. Jena glass filter crucibles (coarse) may be used
RESULTS O F TEST OF METHOD 111-The results obtained in place of t h e Gooch crucibles specified in the directions.
from the analyses of five samples of cellulosic materials by eight (4) Testing for Acidity. T o test the acidity of the wash water on litmus
cooperating laboratories are recorded in the accompanying table, paper, the last few drops removed after the addition of water should be
DEVELOPMENT OF METHODIV-A careful study of the re- used.
sults from the test of Method I11 and of the comments from the
cooperating laboratories led the Committee to believe that the RESULTSOF TESTOF METHODIV-The results obtained from
time allowed for macerating hard pulps should be increased and the analysis of four samples of cellulosicmaterial by nine cooperat-
that only part of the alkali should be added to the sample be- ing laboratories are recorded in the accompanying table.
fore maceration is begun; the remainder added during macera- Some standard cotton cellulose8 was also analyzed according
tion as noted in the directions. The method for standardizing to Method IV for its alpha-cellulose content. Two sets of tripli-
the sodium hydroxide solution was also simplified. cate analyses gave the following results:
Pulp "-600 consisted of a mixture of a medium soft and a hard (1) (2)
pulp. This sample caused the analysts considerable difficulty 99.45 99.60
in macerating it. Cotton C-200 was known to furnish difficulties 99.52 99.70
in alpha-cellulose determination and was also included to give 99.59 99.72
the method a severe test. Av. 99.52 99.67