binder-carbon. This binder-carbon is still combustible in the presence of an excess of coke powder. Owing t o the severe requirements for railway motor brushes, the work for a long time was devoted exclusively t o this field and there is probably still plenty of improvement possible. As the work has advanced, the refinements of requircments have bccome more and more apparent, and i t is quite evident nom that several different carbon brush types are neccssary t o satisfy the requirements for differcnt types of machines. In case of this particular brush, however, it is worth attention that by slight modifications in the process. such as fineness of grinding, pressure on the hydraulic press, etc., thc record of tests kcpt during the past 2 years shows the following changes:
FIG.6.
shrinkage in the two widely different materials, a hard brush body and a body largely composed of graphite. A sort of laminated brush is thereby produced and if there were service demands lor such laminated brushes they could be produced. R~ESEAXCX LBBOKITORY, eLsnxlc eo.,
GEWllhlL
.%TZEyBCI11DY.
THE RELATION OF THE REFRACTIVE INDEX OF SODA. LIME GLASSES TO THEIR CHEMICAL COMPOSITION. By EDWINWARDT~LLOTSON, Jn.
'Through the researches' of Abbe, Schott, Winkelmann and others, i t has been shown that many of the 0.00i20 physical constants of glasses may be roughly calcuAttempts were mad? to operate without change in lated, with the aiil of empirical factors, from. their the components, as weighed out; all qualitics hzve chemical composition. In all ol those calculations i l was assumed that the constant of t h e glass was the improved through small refinements. additive sum of the empirically dctcrmincd constants Devialopment of the railway brush led to trying the same product on other types of electrical apparatus, of the oxides present, and i t was shown that these empirical constants. since they must contain an eleand i t w a s at once evident that the general brush mcnt characteristic of the manner in which the oxide requirements call for more than one kind of brush is combincd in the glass, may be very diffcrcnt from and more than a single composition. Generator the constant of the corresponding pure oxide. No brushes, vhile they (Io not meet the severe conditions data was presented for calculating the refractive met by railway motor brushes and may thcreforc index, although, both from the scientific and indusbe softer and of lower physical tests. should have high conductivity and should in use develop a polished trial standpoints, it is one of the most valnable of constants. commutator without cut,ting or smutting the metal. Many expressions for the relation of the refractive A fairly satisfactory typc of generator brush may he index to the density of a substance have been demade almost entirely of ash-free graphite and bindercarbon and will have approximately the following veloped. but only two have theoretical significance. N-1 values on test: Hardness, 35 ;resistance, 0.00078; tensile The first of these, = I(, %-as proposed* b y d strength, 2 5 0 0 ; chip test, 6 . It has heen found that i a brush needs some lubrication qualities which are Gladstone and Dale and the second, N - - 1 d K, difficult to express quantitatively. In the past it was simultaneously and independently dcveioped has even been customary to treat some brushes with oils, vaseline, etc., to give them this lubricating cficct. by Lorenzj and Lorent%.* In both of these expresIt has also been found possible t o improvc opcration sions I(,the specific refractive power, is a constant of a commutator b y using hard, strong, non-lubricating which is supposedly indepcndent of the temperature, brushes and to interpose in several of the brush- and which is defined b y the chemical composition holders a pure graphite brush, which serves to give and constitution of the substance. Many cxccptions desired lubrication for the other brushes. This has io this rule occur, however, and the expression of led io experiments on a combination brush, of which Gladstone and Dale fails completely whcn the vari2 "TcnnGliirs and itsScientii6c m d Induslrial Ag~liealion."floveitsdr. a section is shown (Fig. 6 ) . I n this case, by a proper Z P h i J . Tronr.. 331, 1861. selection of proportions of binder or pitch and the Wicd. Ann., 11, 70 0 8 8 0 ) . mixture, it has become possible t o gain the same 4 lliid., 9, 611 (1880). Hardners. 1909 . . . . . . . 5 2 . 3
Rcristtnnce.
Tensile strength.
0.00135
1800 2700
for glass prepared in clay crucibles is shown by the fact t h a t , even in the manufacture of optical glass, which must possess homogeneity to the highest degree, it is customary1 to find a difference in a large plate amounting t o several units in the fourth decimal place and not uncommon t o find a difference of one unit in the third decimal place. and also,by the fact t h a t manufacturers cannot vouch2 for the refractive index of a crown glass t o within two units, or for a K =P 2 K + P--y-’K. + 2 PK flint glass to within six or eight units in the third ...... I O 0 IO0 IO0 decimal place. The errors arising from preparation and from optical measurements are therefore probably in which p , , p z , p s , etc., are the percentages in which small in comparison to those introduced into the the several compounds are present, and K,, K,, K3, values of the specific refractivities by the use of the etc., are their respective specific refractivities. calculated densities. I t has been shown in a preVery few attempts have been made to apply this vious papers that the densities uf many glasses may relation t o glasses or t o pure silicates. Larsen’ has be calculated with a moderate degree of accuracy, shown b y some very accurate measurements on carefrom the chemical composition. The average error fully prepared silicates t h a t the specific refractive power is additive within limits of error in case of in these computed densities was about one per cent., but occasionally amounted to two per cent. An mixtures of calcium and magnesium metasilicates, error of one per cent. in the density produces an error and of albite and anorthite, both in the vitreous and the in the refractive index, when i t is about 1.5, of five crystalline form. A marked difference, however, occurs units in the third decimal place, and about one unit in the values of the specific, refractive power for the in the second place when i t is 1.6. In the specific vitreous and crystalline compounds of like composition. refractivity (Lorenz and Lorentz), one per cent. This paper is a n account of an investigation on the error in the density introduces a n error of from 0 . 0 0 0 5 refractive index and the specific refractive power of O.OOIO. I t is also possible that a n additional error some soda lime glasses for the purpose of obtaining was introduced in the composition of the glasses. data, whereby these refractive indices might be cal- No analyses of the glasses were made, but the comculated from their composition with a fair degree of position was assumed to correspond t o that of the accuracy. The glasses were prepared from pure “batch.” This error is doubtless small in comparison sodium carbonate, pure calcium carbonate and a with other errors since Kultascheff4 has shown that, high 2rade of glass sand which analyzed 98. j % S O , , although the loss on heating pure sodium metasilicate the remainder being chiefly aluminum oxide and may amount t o 4 % , as the percentage of calcium moisture. The ingredients mere carefully weighed silicate increases the loss rapidly decreases and beout, well mixed and finally fused in clay crucibles comes negligible; a n d Day and Allen5 apparently in a n oxidizing atmosphere of a gas furnace. When did not observe any noticeable volatilization in the the glass was “plain” i t was‘poured out on cold iron synthesis of feldspar. plates and, after annealing, was broken into small The results which were obtained are given in Table fragments. The refractive index for white light was I. The first four columns of this table shots. the commeasured with the aid of a n AbbC refractometer using position of the glasses, which correspond to mixtures selected fragments of the glass which presented a ranging from Ka,O.gSiO, to CaO.Si0,. I n the fifth smooth, flat surface of the original plate. With a column are given the densities as calculated6 from little care in selecting plane fragments of the glass, the composition : in the sixth, the observed refractive the values of the refractive index for successive plates index for white light; and in the seventh, the values usually agreed t o within two or three units in the of K which were calculated from Gladstone and Dale’s third decimal place, and the average of a number of formula. readings was finally taken. When these values for K are plotted in a system, I t is evident that, in the procedure as set forth, in which the ordinates represent the composition there are several chances for error. The glass could and the abscissas the specific refractivities, as shown not be obtained free from striae in one or even two in Fig. I , i t becomes evident that they form two fusions when clay crucibles were employed, owing to straight lines which intersect a t a point corresponding the solvent action of the glass on the crucible; and the t o a glass of the composition aNaZO.3CaO.gSiO,. melts were stirred very little, therefore, in order t h a t This break in the curve therefore, indicates the exa s few impurities as possible should be introduced. istence of a compound a t this point. I t is especially I t is improbable t h a t the lack of homogeneity, as significant since both Kultascheff 7 and Wallace8 have shown by striae, introduced a very marked error, since the refractive indices of successive Dreparations* Zscholke, Z e d . fur Instrumenfenk, 29, 286. of the same glass, measured on carefully ground and ;ff;&::;3, 897 (1911), 4 Z. anorg. c h e m . , 35, 187 (1903). polished plates, agreed to within two units in the third 5 “ T h e Isomorphism and Thermal Properties of the Feldspars.” decimal place. That this is within the limit of error
ation of temperature is sufficient to produce a change of state in the substance. Both expressions hold fairly well, however, for liquids and solutions. I n a solution or a homogeneous mixture of liquids, which do not react to form a compound, the specific refractive power is a n additive function of the respective refractivities of the compounds present and may therefore be calculated as follows:
+
1
1~
1 2
Am. J. Sci., 28, 263 (1909). Unpublished d a t a .
~
~
Hovestadt, loc. c i t . Tillotson, loc. cit. anavo’. Chem., 36, 187 (1903). I b i d . , 63, 1 (1909).
7 Z.
~
~
+
~
T H E JOUR-Va4L OF I - Y D L 7 S T R I A L A S D E.YGI-\-EERI.VG
248
CHEMISTRY.
April,
1912
TABLEI. CaSiO3 mol. P e r cent.
00 10 20 30 40 50 50 60 70 80 80 100 1 -N,a-Larsen,
SiOz. P e r cent. 74.45 73.60 72.32 70.85 69.30 67.40 67.40 65.20 62 .75 59.60 59.60 51.85 loc czt
CaO. Per cent.
h-azO. Per cent.
D calc.
N obs.
0.00 2.50 5.18 8.25 1l.iO 15.60 15.60 20.20 25.35 31 .SO 31 .SO 48.15
25.55 23.90 22.50 20.90 19.00 17 . O O 1 7 .OO 14.60 11.90 8.60 8.60
2.37 2.395 2.42 2.45 2 48 2 52 2.52 2 5i 2.63 2.705 2.705 2.92
1.5000 1.5060 1.5115 1.5172 1.5240 1.5334 1.5300 1 ,5432 1 ,5590 1.5800 1.5796 1 .62S1
...
observed t h a t in mixtures of sodium and calcium metasilicates a maximum melting point is reached a t the composition aNa,O.gCaO.gSiO,. In column 8
K obs.
K calc.
0.2110 0.2112i 0 . 2 1 137 0 . 2 1 110 0.21129 0 . 2 1166 0.21032 0.21137 0.21254 0.21442 0.21436 0.21517
0.2110 0.2111 0.21118 0.21125 0.21135 0.2115 0.2115 0.2116 0.2127 0.2142 0.2142 0.2178
?:
.\$
calc.
ca1c.-obs.
1 ,5000 1 ,5055 1.5111 1.5175 1.52410 1.5330 1.5330 1.5438 1.5594 1.5794 1.5794 1 ,6307
0 .oooo -0.0005 -0.0004 0.0003 0.0001 -0.0004 f0.0030 f0.0006 4-0.0004 -0 ,0006 -0.0002 +0.0027
+
+
tion of the glass. Since the specific refractivity is. within the limits of error, linear, it may be computed b y means of the following equation:
K
=
pJK1 IO0
+ 'ZK2 I00
1 IO0
, etc.
in which P,, P,, P,, etc., are the percentages of the oxides, and K,, K,, K,. etc., are the empirically determined specific refractivities of the oxides, For soda lime glasses these are as follows:
P..r.nt
SiOa ................................... CaO.. ................................. Nap0 ..................................
i,w t
F.s I
(Table I ) are given the specific refractivities, as indicated b y the two straight lines in Fig. I ; column 9, the refractive, index computed from columns 5 and 8 ; and column I O , the difference between the calculated and observed values of the refractive index. I n Table I1 are shown similar data for these same glasses, calculated. from the formula of Lorentz and Lorenz, and in Fig. 2 the graphic relation of the specific refractivities t o the composition, I t will be noticed t h a t the same break is observed in the refractive curve as shown by the full lines in Fig. 11. TABLE11. CaSiOs mol. Per cent. 00 10 20 30 40 50 50 60 70 80 80 100
D calc. 2.37 2.395 2.42 2.45 2.48 2.52 2.52 2.57 2.63 2.705 2.705 2.92
- -
2.432
s obs.
K' obs.
K' calc.
0.1241 0.12422 1.500 0.12406 0.12389 1.506 0.12349 1.5115 0.1239 1.5172 0.12351 0.12361 0.12338 0.12346 1.524 1 5334 0.12329 0.12323 0.12260 0.12323 1.530 1 ,5432 0.12272 0.12300 0.12277 0.12270 1.559 0.12305 0.12236 1.580 1.5796 0,12298 0.12236 1 ,62S1 0.12155 0.12150
N
s
calc.
calc.-obs.
1.5006 1.5054 1.5110 1.5176 1.5244 1.5333 1.5333 1.5450 1.5586 1,5761 1.5761 1 ,6276
0.0006 -0.0006 -0.0005 0.0004 0.0004 -0.0001 0,0033 +0.0018 -0.0004 -0.0039 -0.0035 -0.0004
- -~ 1.5096
0.12300
0.12291
-
+
+ + +
I ,5093 -0.0003
This break is, however, less marked than in the case shown in Fig. I , and all of the specific refractivities may, for convenience, be considered t o lie upon a single straight line indicated by the dotted line in the figure. This behavior of the Na formula, that of minimizing the effects of compounds on the specific refractivity, makes i t possible to calculate the specific refractivities directly from the percentage composi1
0.1220 0.1210 0.1302
?.'+,a Larsen, loc czt. Si02, 77 7 7 $ , CaO, 8 5 % , NacO, 13 8 %
Columns 5 and 6 of Table I1 show the specific refractivities and refractive indices as calculated b y the 'aid of these factors and column 7 the difference between the calculated and observed values of the refractive index. Not only are the computed values for the glasses given above sufficiently accurate for many kinds of work, but the last glass in Table I1 shows t h a t the refractive index of glasses richer in silica, may also be calculated with equal accuracy. Although, in employing this method for estimating the refractive index, no consideration is taken of the compounds which may be formed, yet it is theoretically correct up to the point where the compound is found, and is applicable t o the majority of soda lime glasses; for these rarely contain as large a proportion
,? z
Lr1140
&m
%,m
t
%LIO
'.", ',
? . :
F 9
z
of lime as the ratio sNa,0.3Ca0.xSI02 calls for, and they therefore lie t o the left (Figs. I and 11). of the break in the specific refractivity curve, and in the region for which the factors, given above, furnish the most satisfactory agreement. I t must be remembered, however, t h a t the factors employed represent not the true specific refractivity of the pure oxide, but the true refractivity modified by the nature of the silicate which is formed and also by the inexact values of the density which is employed in the calculations. This is well illustrated in Table 111, in which are shown therefractive indices of several oxides, together with the density observed. the density em-
ployed in the computations and the specific refractivities calculated from them, together with the factors given above. TABLE 111. D obs.
N. Quartz.. .. . . Trpdymite.. Fused Quartz
CaO.,
.. . . . .
,
. ... .
D calc.
K obs.
P
K3.
calc.?
(0.119611 n.1380] 1.48301 2.31811 2.3 { 0.123191 0.1240 )0.1220 1.45901 2.2131) [,0.1235410 . l l S S J 1.83201 3.3161 4.1 0.13266l 0.1073 0.1210
1 ,5472‘ 2 .65Z1
SUMXARP. I . A series of soda lime glasses have been made and their refractive indices measured. 2 . The existence of a double silicate of the type zNa,0.3Ca0.xSi02, which is doubtless the double metasilicate, described by Kultascheff and Wallace, has been made evident. 3. The specific refractivities of these glasses, computed with the aid of the calculated densities, are additive from pure sodium silicate up to the composition in which the molecular proportion of soda t o lime is z : 3. 4. Factors have been derived b y means of which the specific refractivity of soda lime glasses may be calculated. These are for the formula of Lorentz and Lorenz: S O , , 1 2 2 0 ; CaO, 1 2 1 0 ; Na,O, 0.1302.
VNlVERSITY
OF
KAXSAS,
I A W R E N C E , KANSAS
A NEW METHOD F O R THE DETERMINATION O F VANADIUM. B y D. J. DEMOREST.
Received September 25, 1911
The following method for the determination of vanadium in steel depends upon the selective oxidation of ferrous sulfate in the presence of vanadyl sulfate b y means of manganese dioxide. The vanadyl sulfate is then titrated b y adding an excess of permanganate, the excess permanganate being titrated b y sodium arsenite. This differential oxidizing action apparently contradicts the results of J. R. Cain,4 who found that both iron and vanadium are oxidized, but the reasons for this discrepancy are shown in the note which follows on page 2 5 6 . The manganese dioxide should be sufficiently fine t o pass through a zoo-mesh sieve, and yet should settle in a beaker of water in 30 seconds. The process in detail is as follows: In a joo cc. flask a two-gram sample of the steel or iro? is dissolved in a mixture of 30 cc. of water a n a 1 2 cc. concentrated H,SO, with application of heat. Then one cc. of H N O , (sp. gr. I . 4 z ) is added cautiously to oxidize the iron and the solution is boiled for a few minutes to remove the nitrous fumes. Then the solution is diluted with 30 cc. of water and a strong solution of KMnO, is added to completely oxidize all carbon, etc., and the solution is boiled. If the perLarsen, lac. cit. 2 Using Lorentz and Lorenz formula and “ D calc.” 3 Factors used in computing specific refractivities. 4 T ~ r JsO U R N A L3, , 476 (1911). 1
manganate or the resulting Mr10, should disappear, not enough permanganate has been used. and more should be added. Now ferrous sulfate is added t o reduce the Rho,, HMnO,, H,CrO,, and H,VO,, etc., and the solution is again boiled t o remove any possible nitrous fumes. Then pure distilled water is added to make the volume about 2 5 0 CC., N / I O KXnO, added until the solution is pink, and the solution cooled t o t a p water temperature. Ferrous sulfate solution is added until all reducible compounds including chromic and vanadic acids are reduced. Only enough ferrous sulfate should be added t o be certain that there is a decided excess present. A solution, one cc. of which equals about 0 . 0 1 gram of iron, is the one used, Now about one gram of C. P. MnO, is added and the solution shaken vigorously. After two minutes a drop is tested with ferricyanide on a white plate t o see if the iron is completely oxidized. I t generally takes from four t o six minutes. At the end of each minute the solution is tested for ferrous iron until none is present and the shaking is continued for about one-half minute longer. I t should be noted t h a t a bluish color will always be obtained in the presence of vanadyl sulfate after the test drop has stood for a few seconds. The end should be taken when the test does not show blue immediately. The blue color which forms after a few seconds, even when there is no ferrous iron present, is due to the reduction of ferri- to ferrocyanide by the vanadyl sulfate. One can become familiar with this end b y adding a drop of ferric sulfate containing vanadyl sulfate t o a drop of ferricyanide on a white plate. The MnO, oxidizes the ferrous sulfate t o ferric sulfate, but does not oxidize the vanadyl sulfate [V2O2(SO4)J. Then the MnO, is filtered off on an asbestos mat, using suction. From a burette a standard solution of KMnO, is added until a pink tinge is present in the solution, and one cc. more is added, and after one minute the excess permanganate is titrated with Na,AsO, solution. The end point is very sharp. If a t this point the operator is not satisfied with this titration, the excess arsenite may be oxidized with KkInO,, ferrous sulfate again added, then oxidized with MnO, as before, arid the titration repeated, thus giving a check on the titration. A blank determination must be run on a vanadium-free steel, and the result deducted. The blank generally amounts t o about 0.00075 gram V. The time required is about one-half hour and the results are very satisfactory. I n fact the accuracy is about that of a phosphorus determination, The vanadium steel standard furnished b y the Bureau of Standards mas analq-zed by the above method. The result of the Bureau chemists is 0.0143 per cent. V and the average of the cooperating chemists is 0 . I j per cent. V. The writer obtains the following results, the average being-’o.I43 per cent. : 0.140 0.147 0.143
0.138 0.147 0,143
To further test the method, two-gram samples of
